Business Understanding

The project aims to improve object detection in optical remote sensing images using an adaptive Mask RCNN model. This technology is critical for applications in both civilian and military domains, where fast and accurate identification of objects in satellite imagery is required.

Surveillance and Security:

• Detection of unauthorized vehicles or aircraft in restricted areas.
• Monitoring of military bases or borders to detect potential threats.
• Identifying unusual activities in high-security zones.

Urban Planning and Infrastructure:

• Detecting changes in urban environments, such as new construction, road networks, and public infrastructure.
• Monitoring the growth of cities and identifying illegal constructions.

Environmental Monitoring:

• Detecting changes in natural landscapes, such as deforestation or water level changes.
• Monitoring environmental disasters like oil spills or forest fires.

Surveillance and Security:

• Detection of unauthorized vehicles or aircraft in restricted areas.
• Monitoring of military bases or borders to detect potential threats.
• Identifying unusual activities in high-security zones.

Load Data ( Positive images / Negative images /Annotation )

from google.colab import drive
drive.mount('/content/drive')

Mounted at /content/drive

import os
import numpy as np
import matplotlib.pyplot as plt
from tensorflow.keras.preprocessing.image import ImageDataGenerator
import cv2

def load_images_from_folder(folder):
    images = []
    filenames = []
    original_sizes = []
    for filename in os.listdir(folder):
        img = cv2.imread(os.path.join(folder, filename))
        if img is not None:
            original_size = img.shape[:2]  # Capture original size
            original_sizes.append(original_size)
            images.append(img)
            filenames.append(filename)
    return images, filenames, original_sizes

def load_annotations_from_folder(folder, filenames, target_size=(224, 224)):
    annotations = []
    for idx, filename in enumerate(filenames):
        annot_filename = os.path.splitext(filename)[0] + '.txt'
        annot_file = os.path.join(folder, annot_filename)
        boxes = []
        with open(annot_file, 'r') as file:
            for line in file.readlines():
                parts = line.replace('(', '').replace(')', '').replace(',', ' ').split()
                i = 0
                while i < len(parts) - 4:
                    try:
                        x_min = int(float(parts[i].strip()))
                        y_min = int(float(parts[i+1].strip()))
                        x_max = int(float(parts[i+2].strip()))
                        y_max = int(float(parts[i+3].strip()))
                        class_id = int(parts[i+4].strip())
                        boxes.append([x_min, y_min, x_max, y_max, class_id])
                        i += 10
                    except ValueError as e:
                        print(f"Error parsing line '{line}': {e}")
                        i += 10
        annotations.append(boxes)
    return annotations

Data Understanding

Overview:

The NWPU VHR-10 dataset comprises images extracted from Google Earth, one of the most widely used platforms for high-resolution satellite imagery. The dataset was manually annotated by experts to ensure accuracy.

Content:

Classes: The dataset contains 10 classes of objects, which include both man-made and natural objects commonly found in remote sensing imagery. The classes are:

Airplane

Ship

Storage Tank

Baseball Diamond

Tennis Court

Basketball Court

Ground Track Field

Harbor

Bridge

Vehicle

Total Images: The NWPU VHR-10 dataset contains a total of 800 images.

Annotations: The dataset includes annotated bounding boxes for each of the 10 object classes. These annotations identify the specific regions within each image where the respective objects are located.

## Suppress warnings for a cleaner output
import warnings
warnings.filterwarnings('ignore')
# Core libraries for data processing and mathematical operations
import matplotlib.pyplot as plt
import numpy as np
# Deep learning libraries
import tensorflow as tf
import tensorflow_hub as hub
from tensorflow.keras import layers
import json
from PIL import Image, ImageDraw
import skimage.draw

from mrcnn.visualize import display_instances, display_top_masks
from mrcnn.utils import extract_bboxes

from mrcnn.utils import Dataset
from matplotlib import pyplot as plt

from mrcnn.config import Config



from mrcnn import utils

WARNING:tensorflow:From C:\Users\Mega-PC\anaconda3\lib\site-packages\tf_keras\src\losses.py:2976: The name tf.losses.sparse_softmax_cross_entropy is deprecated. Please use tf.compat.v1.losses.sparse_softmax_cross_entropy instead.

import os

# Walk through NWPU directory and list number of files
for dirpath, dirnames, filenames in os.walk(r".\NWPU VHR-10 dataset"):
    print(f"There are {len(dirnames)} directories and {len(filenames)} images in '{dirpath}'.")

There are 3 directories and 1 images in '.\NWPU VHR-10 dataset'.
There are 0 directories and 650 images in '.\NWPU VHR-10 dataset\ground truth'.
There are 0 directories and 150 images in '.\NWPU VHR-10 dataset\negative image set'.
There are 0 directories and 650 images in '.\NWPU VHR-10 dataset\positive image set'.

COCO JSON FORMAT

positive_image_folder = r".\NWPU VHR-10 dataset\positive image set"
ground_truth_folder = r".\NWPU VHR-10 dataset\ground truth"

# Define output annotation file
output_annotation_file = "annotations.json"
# Define the mapping of class names to IDs
class_mapping = {
    "airplane": 1,
    "ship": 2,
    "storage tank": 3,
    "baseball diamond": 4,
    "tennis court": 5,
    "basketball court": 6,
    "ground track field": 7,
    "harbor": 8,
    "bridge": 9,
    "vehicle": 10
}
image_id = 0

coco_data = {"images": [], "annotations": [], "categories": []}

for filename in os.listdir(ground_truth_folder):
    if filename.endswith('.txt'):
        txt_file_path = os.path.join(ground_truth_folder, filename)
        image_file_path = os.path.join(positive_image_folder, filename.replace('.txt', '.jpg'))

        with open(txt_file_path, 'r') as f:
            lines = f.readlines()

            # Open the image to get its height and width
            with Image.open(image_file_path) as img:
                image_width, image_height = img.size

            # Add image entry to coco_data
            image_entry = {
                "id": image_id,
                "file_name": image_file_path,
                "height": image_height,
                "width": image_width
            }
            coco_data['images'].append(image_entry)
            image_id += 1

            # Iterate over each line in the text file
            for line in lines:
                line = line.strip()  # Remove leading/trailing white spaces
                # Extract the bounding box coordinates and object class
                values = line.split('),')
                # Ensure the line contains at least 2 values (x1,y1),(x2,y2),a
                if len(values) >= 3:
                    # Extract the values for bounding box coordinates and object class
                    x1, y1 = map(int, values[0].replace('(', '').split(','))
                    x2, y2 = map(int, values[1].replace('(', '').split(','))

                    obj_class = int(values[2])

                    # Create a dictionary for the annotation
                    annotation = {
                        'image_id': image_id - 1,  # Image ID corresponds to index in the images list
                        'category_id': obj_class,  # Assuming object class is the category ID
                        'bbox': [x1,y1,x2,y2],  # COCO bbox format: [x, y, width, height]
                        'area': (x2 - x1) * (y2 - y1),  # Assuming area is bbox width * height
                        'iscrowd': 0  # Set to 0 for non-crowd annotations
                    }

                    # Append the annotation to the coco_data
                    coco_data['annotations'].append(annotation)

# Define categories
categories = [{"id": class_id, "name": class_name} for class_name, class_id in class_mapping.items()]

# Add categories to coco_data
coco_data['categories'] = categories

# Write the coco_data to the output annotation file
with open(output_annotation_file, 'w') as f:
    json.dump(coco_data, f, indent=4)

class CocoLikeDataset(utils.Dataset):
    def load_data(self,annotation_json,images_dir):
        json_file=open(annotation_json)
        coco_json=json.load(json_file)
        json_file.close()
        source_name="coco_like"
        for category in coco_json['categories']:
            class_id=category['id']
            class_name=category['name']
            if class_id<1:
                print('Error: Class id for "{}" cannot be less than one '.format(class_name))
                return
            self.add_class(source_name,class_id,class_name)
        annotations={}
        for annotation in coco_json['annotations']:
            image_id = annotation['image_id']
            if image_id not in annotations:
                annotations[image_id]=[]
            annotations[image_id].append(annotation)


        seen_images={}
        for image in coco_json['images']:
            image_id=image['id']
            if image_id in seen_images:
                print("Warning: Skipping duplicate image id : {}".format(image))
            else:
                seen_images[image_id]=image
                try:
                    image_file_name=image['file_name']
                    image_width = image['width']
                    image_height = image['height']
                except KeyError as key:
                    print("Warning: Skipping image (id: {}) with missing key : {}".format(image_id,key))
                image_path=os.path.abspath(os.path.join(images_dir,image_file_name))
                image_annotations=annotations[image_id]
                self.add_image(
                    source=source_name,
                    image_id=image_id,
                    path=image_path,
                    width=image_width,
                    height=image_height,
                    annotations=image_annotations
                )
    def load_mask(self,image_id):
        image_info = self.image_info[image_id]
        annotations = image_info['annotations']
        instance_masks = []
        class_ids = []

        for annotation in annotations:
            class_id = annotation['category_id']
            mask = Image.new('1', (image_info['width'], image_info['height']))
            mask_draw = ImageDraw.ImageDraw(mask, '1')
            print(annotation['bbox'])
            array=annotation['bbox']
            mask_draw.rectangle(array, fill=1)
            bool_array = np.array(mask) > 0
            instance_masks.append(bool_array)
            class_ids=np.append(class_ids,class_id)

            mask = np.dstack(instance_masks)
            class_ids = np.array(class_ids, dtype=np.int32)

        return mask, class_ids

Resizing of the images

## Suppress warnings for a cleaner output
import warnings
warnings.filterwarnings('ignore')
# Core libraries for data processing and mathematical operations
import matplotlib.pyplot as plt
import numpy as np
# Deep learning libraries
import tensorflow as tf
import tensorflow_hub as hub
from tensorflow.keras import layers
import json
from PIL import Image, ImageDraw
import skimage.draw

from mrcnn.visualize import display_instances, display_top_masks
from mrcnn.utils import extract_bboxes

from mrcnn.utils import Dataset
from matplotlib import pyplot as plt

from mrcnn.config import Config



from mrcnn import utils

WARNING:tensorflow:From C:\Users\Mega-PC\anaconda3\lib\site-packages\tf_keras\src\losses.py:2976: The name tf.losses.sparse_softmax_cross_entropy is deprecated. Please use tf.compat.v1.losses.sparse_softmax_cross_entropy instead.

import os

# Walk through NWPU directory and list number of files
for dirpath, dirnames, filenames in os.walk(r".\NWPU VHR-10 dataset"):
    print(f"There are {len(dirnames)} directories and {len(filenames)} images in '{dirpath}'.")

There are 3 directories and 1 images in '.\NWPU VHR-10 dataset'.
There are 0 directories and 650 images in '.\NWPU VHR-10 dataset\ground truth'.
There are 0 directories and 150 images in '.\NWPU VHR-10 dataset\negative image set'.
There are 0 directories and 650 images in '.\NWPU VHR-10 dataset\positive image set'.

positive_image_folder = r".\NWPU VHR-10 dataset\positive image set"
ground_truth_folder = r".\NWPU VHR-10 dataset\ground truth"

# Define output annotation file
output_annotation_file = "annotations.json"
# Define the mapping of class names to IDs
class_mapping = {
    "airplane": 1,
    "ship": 2,
    "storage tank": 3,
    "baseball diamond": 4,
    "tennis court": 5,
    "basketball court": 6,
    "ground track field": 7,
    "harbor": 8,
    "bridge": 9,
    "vehicle": 10
}
image_id = 0

coco_data = {"images": [], "annotations": [], "categories": []}

for filename in os.listdir(ground_truth_folder):
    if filename.endswith('.txt'):
        txt_file_path = os.path.join(ground_truth_folder, filename)
        image_file_path = os.path.join(positive_image_folder, filename.replace('.txt', '.jpg'))

        with open(txt_file_path, 'r') as f:
            lines = f.readlines()

            # Open the image to get its height and width
            with Image.open(image_file_path) as img:
                image_width, image_height = img.size

            # Add image entry to coco_data
            image_entry = {
                "id": image_id,
                "file_name": image_file_path,
                "height": image_height,
                "width": image_width
            }
            coco_data['images'].append(image_entry)
            image_id += 1

            # Iterate over each line in the text file
            for line in lines:
                line = line.strip()  # Remove leading/trailing white spaces
                # Extract the bounding box coordinates and object class
                values = line.split('),')
                # Ensure the line contains at least 2 values (x1,y1),(x2,y2),a
                if len(values) >= 3:
                    # Extract the values for bounding box coordinates and object class
                    x1, y1 = map(int, values[0].replace('(', '').split(','))
                    x2, y2 = map(int, values[1].replace('(', '').split(','))

                    obj_class = int(values[2])

                    # Create a dictionary for the annotation
                    annotation = {
                        'image_id': image_id - 1,  # Image ID corresponds to index in the images list
                        'category_id': obj_class,  # Assuming object class is the category ID
                        'bbox': [x1,y1,x2,y2],  # COCO bbox format: [x, y, width, height]
                        'area': (x2 - x1) * (y2 - y1),  # Assuming area is bbox width * height
                        'iscrowd': 0  # Set to 0 for non-crowd annotations
                    }

                    # Append the annotation to the coco_data
                    coco_data['annotations'].append(annotation)

# Define categories
categories = [{"id": class_id, "name": class_name} for class_name, class_id in class_mapping.items()]

# Add categories to coco_data
coco_data['categories'] = categories

# Write the coco_data to the output annotation file
with open(output_annotation_file, 'w') as f:
    json.dump(coco_data, f, indent=4)

class CocoLikeDataset(utils.Dataset):
    def load_data(self,annotation_json,images_dir):
        json_file=open(annotation_json)
        coco_json=json.load(json_file)
        json_file.close()
        source_name="coco_like"
        for category in coco_json['categories']:
            class_id=category['id']
            class_name=category['name']
            if class_id<1:
                print('Error: Class id for "{}" cannot be less than one '.format(class_name))
                return
            self.add_class(source_name,class_id,class_name)
        annotations={}
        for annotation in coco_json['annotations']:
            image_id = annotation['image_id']
            if image_id not in annotations:
                annotations[image_id]=[]
            annotations[image_id].append(annotation)


        seen_images={}
        for image in coco_json['images']:
            image_id=image['id']
            if image_id in seen_images:
                print("Warning: Skipping duplicate image id : {}".format(image))
            else:
                seen_images[image_id]=image
                try:
                    image_file_name=image['file_name']
                    image_width = image['width']
                    image_height = image['height']
                except KeyError as key:
                    print("Warning: Skipping image (id: {}) with missing key : {}".format(image_id,key))
                image_path=os.path.abspath(os.path.join(images_dir,image_file_name))
                image_annotations=annotations[image_id]
                self.add_image(
                    source=source_name,
                    image_id=image_id,
                    path=image_path,
                    width=image_width,
                    height=image_height,
                    annotations=image_annotations
                )
    def load_mask(self,image_id):
        image_info = self.image_info[image_id]
        annotations = image_info['annotations']
        instance_masks = []
        class_ids = []

        for annotation in annotations:
            class_id = annotation['category_id']
            mask = Image.new('1', (image_info['width'], image_info['height']))
            mask_draw = ImageDraw.ImageDraw(mask, '1')
            print(annotation['bbox'])
            array=annotation['bbox']
            mask_draw.rectangle(array, fill=1)
            bool_array = np.array(mask) > 0
            instance_masks.append(bool_array)
            class_ids=np.append(class_ids,class_id)

            mask = np.dstack(instance_masks)
            class_ids = np.array(class_ids, dtype=np.int32)

        return mask, class_ids

LOADING A COCOJSON DATA

image_id = 1
# load the image
image = dataset_train.load_image(image_id)
# load the masks and the class ids
mask, class_ids = dataset_train.load_mask(image_id)

# display_instances(image, r1['rois'], r1['masks'], r1['class_ids'],
# dataset.class_names, r1['scores'], ax=ax, title="Predictions1")

# extract bounding boxes from the masks
bbox = extract_bboxes(mask)
# display image with masks and bounding boxes
display_instances(image, bbox, mask, class_ids, dataset_train.class_names)

[575, 114, 635, 162]
[72, 305, 133, 369]
[210, 317, 273, 384]
[306, 374, 344, 420]
[447, 531, 535, 632]
[546, 605, 625, 707]
[632, 680, 720, 790]

import cv2

def resize_images(images, target_size=(1024, 1024)):
    resized_images = []
    for img in images:
        resized_img = cv2.resize(img, target_size)
        resized_images.append(resized_img)
    return resized_images

def resize_annotations(annotations, original_sizes, target_size=(1024, 1024)):
    resized_annotations = []
    for idx, annot in enumerate(annotations):
        original_height, original_width = original_sizes[idx]
        scale_x = target_size[0] / original_width
        scale_y = target_size[1] / original_height
        resized_boxes = []
        for box in annot:
            x_min = int(box[0] * scale_x)
            y_min = int(box[1] * scale_y)
            x_max = int(box[2] * scale_x)
            y_max = int(box[3] * scale_y)
            class_id = box[4]
            resized_boxes.append([x_min, y_min, x_max, y_max, class_id])
        resized_annotations.append(resized_boxes)
    return resized_annotations

# Resizing positive images
target_size = (1024, 1024)  # Define your target size here
resized_positive_images = resize_images(positive_images, target_size)
resized_negative_images = resize_images(negative_images, target_size)

# Resizing positive annotations
resized_positive_annotations = resize_annotations(positive_annotations, positive_original_sizes, target_size)

Data augmentation

Multiple Individual Augmentations:

Pros: This method allows the model to learn from a wider variety of simple changes, potentially improving its ability to generalize from each type of transformation individually. It increases the effective size of the training dataset more significantly, which can be beneficial for training deep learning models.

Cons: The main drawback is increased computational and storage requirements since you're generating multiple images for each original image in the dataset. It might also introduce redundancy if the transformations are too mild or too correlated. Best Practices

Balanced Approach: Often, a mix of both strategies is employed. For example, you might apply mild transformations (like slight rotations and flips) individually to generate multiple images and then perform a few combined transformations (like moderate zoom followed by a slight rotation) to create more diverse scenarios. Experimentation and Validation: It's important to experiment with different strategies and validate their impact on model performance. Monitoring how each type of augmentation affects overfitting, underfitting, and validation accuracy can guide you to optimize the augmentation pipeline. Resource Management: Consider your computational resources and training time. More images mean longer training times and more disk space. If resources are limited, focusing on the most impactful transformations might be necessary.

Data Augmentation For Negative Images

import numpy as np
import cv2
import matplotlib.pyplot as plt

def adjust_brightness_negative(image, brightness_factor):
    """ Adjust the brightness of an image. """
    return cv2.convertScaleAbs(image, alpha=brightness_factor, beta=0)

def flip_horizontal_negative(image):
    """Flip image horizontally."""
    return cv2.flip(image, 1)  # 1 means horizontal flip

def flip_vertical_negative(image):
    """Flip image vertically."""
    return cv2.flip(image, 0)  # 0 means vertical flip

def generate_augmented_images_negative(original_img, zoom_factors, angles, brightness_factors):
    augmented_images = []

    # 1. Zoom + Rotation:
    zoom_factor = np.random.choice(zoom_factors)
    new_width = int(original_img.shape[1] * zoom_factor)
    new_height = int(original_img.shape[0] * zoom_factor)
    zoomed_rotation_img = cv2.resize(original_img, (new_width, new_height))
    start_x = (new_width - original_img.shape[1]) // 2
    start_y = (new_height - original_img.shape[0]) // 2
    zoomed_rotation_img = zoomed_rotation_img[start_y:start_y + original_img.shape[0], start_x:start_x + original_img.shape[1]]
    angle = np.random.choice(angles)
    zoomed_rotation_img = cv2.warpAffine(zoomed_rotation_img, cv2.getRotationMatrix2D((zoomed_rotation_img.shape[1] / 2, zoomed_rotation_img.shape[0] / 2), angle, 1), (zoomed_rotation_img.shape[1], zoomed_rotation_img.shape[0]))
    augmented_images.append(zoomed_rotation_img)

    # 2. Brightness Adjustment + Vertical Flip:
    v_flip_img = flip_vertical_negative(original_img)
    brightness_factor = np.random.choice(brightness_factors)
    v_flip_img = adjust_brightness_negative(v_flip_img, brightness_factor)
    augmented_images.append(v_flip_img)

    # 3. Zoom + Brightness Adjustment:
    zoom_factor = np.random.choice(zoom_factors)
    new_width = int(original_img.shape[1] * zoom_factor)
    new_height = int(original_img.shape[0] * zoom_factor)
    zoomed_brightness_img = cv2.resize(original_img, (new_width, new_height))
    start_x = (new_width - original_img.shape[1]) // 2
    start_y = (new_height - original_img.shape[0]) // 2
    zoomed_brightness_img = zoomed_brightness_img[start_y:start_y + original_img.shape[0], start_x:start_x + original_img.shape[1]]
    brightness_factor = np.random.choice(brightness_factors)
    zoomed_brightness_img = adjust_brightness_negative(zoomed_brightness_img, brightness_factor)
    augmented_images.append(zoomed_brightness_img)

    # Rotation augmentation
    angle = np.random.choice(angles)
    rotated_img = cv2.warpAffine(original_img, cv2.getRotationMatrix2D((original_img.shape[1] / 2, original_img.shape[0] / 2), angle, 1), (original_img.shape[1], original_img.shape[0]))
    augmented_images.append(rotated_img)

    # Horizontal Flip
    h_flip_img = flip_horizontal_negative(original_img)
    augmented_images.append(h_flip_img)

    return augmented_images

def display_augmentations_negative(original_img):
    zoom_factors = [1.2, 1.4, 1.6, 1.8, 2.0]  # Factors for zooming
    angles = [10, 20, 30, -10, -20, -30]       # Degrees for rotation
    brightness_factors = [0.5, 0.7, 1.3, 1.5]  # Factors for brightness adjustment

    aug_imgs = generate_augmented_images_negative(original_img, zoom_factors, angles, brightness_factors)
    fig, axes = plt.subplots(1, len(aug_imgs), figsize=(20, 5))  # Adding original image as well
    axes[0].imshow(cv2.cvtColor(original_img, cv2.COLOR_BGR2RGB))
    axes[0].set_title("Original")
    axes[0].axis('off')

    for ax, img in zip(axes[1:], aug_imgs):
        ax.imshow(cv2.cvtColor(img, cv2.COLOR_BGR2RGB))
        ax.axis('off')

    plt.tight_layout()
    plt.show()

# Example usage (assuming you have resized_negative_images defined)
for img in resized_negative_images[:1]:  # Displaying first 5 images for example
    display_augmentations_negative(img)

import numpy as np
import matplotlib.pyplot as plt
import cv2  # Ensure OpenCV is imported for image processing

# Initialize lists to hold all negative images
all_negative_images = []

# Iterate over each negative image
for img in resized_negative_images:

    zoom_factors = [1.2, 1.4, 1.6, 1.8, 2.0]  # Factors for zooming
    angles = [10, 20, 30, -10, -20, -30]       # Degrees for rotation
    brightness_factors = [0.5, 0.7, 1.3, 1.5]  # Factors for brightness adjustment
    # Generate augmented images
    augmented_images = generate_augmented_images_negative(img, zoom_factors, angles, brightness_factors)

    # Append the original image first
    all_negative_images.append(img)

    # Extend list with the augmented images
    all_negative_images.extend(augmented_images)

# Displaying the original and augmented images
fig, axes = plt.subplots(2, 4, figsize=(20, 8))  # Adjust the subplot grid as necessary
for ax, img in zip(axes.flatten(), all_negative_images[:8]):  # Ensure we don't exceed the grid size
    ax.imshow(cv2.cvtColor(img, cv2.COLOR_BGR2RGB))  # Convert BGR to RGB
    ax.axis('off')
plt.tight_layout()
plt.show()

# Calculate the number of images in all_images
number_of_negative_images = len(all_negative_images)

# Print the result
print("Total number of images Negative images after data augmentation:", number_of_negative_images)

MODELING

Fine Tuning ResNet 50

# Download TorchVision repo to use some files from
# references/detection
!pip install pycocotools --quiet
!git clone https://github.com/pytorch/vision.git
!git checkout v0.3.0

!cp vision/references/detection/utils.py ./
!cp vision/references/detection/transforms.py ./
!cp vision/references/detection/coco_eval.py ./
!cp vision/references/detection/engine.py ./
!cp vision/references/detection/coco_utils.py ./

Cloning into 'vision'...
remote: Enumerating objects: 502456, done.
remote: Counting objects: 100% (16066/16066), done.
remote: Compressing objects: 100% (809/809), done.
remote: Total 502456 (delta 15275), reused 15996 (delta 15230), pack-reused 486390
Receiving objects: 100% (502456/502456), 973.34 MiB | 23.03 MiB/s, done.
Resolving deltas: 100% (468510/468510), done.
fatal: not a git repository (or any of the parent directories): .git

import torch
import torchvision
from torch.utils.data import Dataset, DataLoader
import torchvision.transforms as transforms
from torchvision.models.detection import FasterRCNN
from torchvision.models.detection.rpn import AnchorGenerator
from torchvision.models.detection.faster_rcnn import FastRCNNPredictor
from PIL import Image
import numpy as np
# Basic python and ML Libraries
import os
import random
import numpy as np
import pandas as pd
# for ignoring warnings
import warnings
warnings.filterwarnings('ignore')

# We will be reading images using OpenCV
import cv2

# xml library for parsing xml files
from xml.etree import ElementTree as et

# matplotlib for visualization
import matplotlib.pyplot as plt
import matplotlib.patches as patches

# torchvision libraries
import torch
import torchvision
from torchvision import transforms as torchtrans
from torchvision.models.detection.faster_rcnn import FastRCNNPredictor

# these are the helper libraries imported.
from engine import train_one_epoch, evaluate
import utils
import transforms as T

# for image augmentations
import albumentations as A
from albumentations.pytorch.transforms import ToTensorV2
from torch.cuda.amp import GradScaler, autocast

scaler = GradScaler()

from torchvision.transforms import functional as F

class ImagesDataset(torch.utils.data.Dataset):

    def __init__(self, files_dir, width, height,ID,CLASS,BOX, transforms=None):
        self.transforms = transforms
        self.files_dir = files_dir
        self.bbox_dir= bbox_dir
        self.height = height
        self.width = width
        # sorting the images for consistency
        # To get images, the extension of the filename is checked to be jpg
        self.imgs = ID
        self.box=BOX
        self.Class=CLASS


        # classes: 0 index is reserved for background
        self.classes= [_,1,2,3,4,5,6,7,8,9,10]
    def __getitem__(self, idx):
        img_name = self.imgs[idx]
        image_path = os.path.join(self.files_dir, img_name +'.jpg')

        # reading the images and converting them to correct size and color
        img = cv2.imread(image_path)
        img_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB).astype(np.float32)
        img_res = cv2.resize(img_rgb, (self.width, self.height), cv2.INTER_AREA)
        #diving by 255
        img_res /= 255.0

        # annotation file
        annot_filename = img_name + '.txt'
        annot_file_path = os.path.join(self.bbox_dir, annot_filename)
        bb=[]
        # cv2 image gives size as height x width
        wt = img.shape[1]
        ht = img.shape[0]
        # convert boxes into a torch.Tensor
        bbox=self.box[idx]
        for box in bbox:
            xmin_corr = (box[0]/wt)*self.width
            xmax_corr = (box[2]/wt)*self.width
            ymin_corr = (box[1]/ht)*self.height
            ymax_corr = (box[3]/ht)*self.height
            bb.append([xmin_corr, ymin_corr, xmax_corr, ymax_corr])
        boxes = torch.as_tensor(bb, dtype=torch.float32)
#         boxes = torch.as_tensor([xmin_corr, ymin_corr, xmax_corr, ymax_corr], dtype=torch.float32)

        # getting the areas of the boxes
        area = (boxes[:, 3] - boxes[:, 1]) * (boxes[:, 2] - boxes[:, 0])
        #area = (boxes[3] - boxes[1]) * (boxes[2] - boxes[0])
        # suppose all instances are not crowd
        iscrowd = torch.zeros((boxes.shape[0],), dtype=torch.int64)

        labels = torch.as_tensor(self.Class[idx], dtype=torch.int64)

        target = {}
        target["boxes"] = boxes
        target["labels"] = labels
        target["area"] = area
        target["iscrowd"] = iscrowd
        # image_id
        image_id=idx
#         image_id = torch.tensor([idx])
        target["image_id"] = image_id


        if self.transforms:
#             print(img_res.shape)
#             print(target['boxes'])
#             print(labels.view(-1))
            sample = self.transforms(image = img_res,
                                     bboxes = target['boxes'],
                                     labels = labels.view(-1))

            img_res = sample['image']
#             img_res = img_res.permute(1, 2, 0)
            target['boxes'] = torch.Tensor(sample['bboxes'])
            target["labels"] = labels.view(-1)
#             target['boxes'] = sample['bboxes']
#             print("---------------------------")
#             print(img_res.shape)
#             print(target['boxes'])
#             print(labels.view(-1))
#             print("----------------------------")
#         print(target)
        return torch.tensor(img_res), target

    def __len__(self):
        return len(self.imgs)

# Send train=True fro training transforms and False for val/test transforms
def get_transform(train):

    if train:
        return A.Compose([
                            A.HorizontalFlip(0.5),
                     # ToTensorV2 converts image to pytorch tensor without div by 255
                            ToTensorV2(p=1.0)
                        ], bbox_params={'format': 'pascal_voc', 'label_fields': ['labels']})
    else:
        return A.Compose([
                            ToTensorV2(p=1.0)
                        ], bbox_params={'format': 'pascal_voc', 'label_fields': ['labels']})

# defining the files directory and testing directory
files_dir = '/content/drive/MyDrive/dataset/positive image set'
neg_dir = '/content/drive/MyDrive/dataset/negative image set'
bbox_dir = '/content/drive/MyDrive/dataset/ground truth'
#random_dir ='/kaggle/input/random-test'

def getressources():
        # Chemin vers votre répertoire contenant les fichiers
        repertoire = "/content/drive/MyDrive/dataset/ground truth"

        # Initialisation des listes
        noms_fichiers = []
        box_data = []
        class_data = []

        # Parcours des fichiers dans le répertoire
        for nom_fichier in os.listdir(repertoire):
            chemin_fichier = os.path.join(repertoire, nom_fichier)
            # Vérification que le chemin correspond à un fichier et non à un répertoire
            if os.path.isfile(chemin_fichier):
                nom_sans_extension = os.path.splitext(nom_fichier)[0]
                with open(chemin_fichier, 'r') as f:
                    contenu_fichier = f.read()
                    noms_fichiers.append(nom_sans_extension)
                    lignes = contenu_fichier.strip().split('\n')
                    boxes = []
                    classes = []
                    for ligne in lignes:
                        elements = ligne.split(',')
                        box = [int(elem.strip("() ")) for elem in elements[:4]]
                        classe = int(elements[4])
                        boxes.append(box)
                        classes.append(classe)
                    box_data.append(boxes)
                    class_data.append(classes)
        return(noms_fichiers,box_data,class_data)

ID,BOX,CLASS=getressources()

print(len(ID))

650

# Transformation for image resizing and normalization
transform = transforms.Compose([
    transforms.Resize((1024, 1024)),
    transforms.ToTensor(),
    transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
])


# use our dataset and defined transformations
dataset = ImagesDataset(files_dir, 480, 480,ID,CLASS,BOX,transforms= get_transform(train=True))
dataset_test = ImagesDataset(files_dir, 480, 480,ID,CLASS,BOX,transforms= get_transform(train=False))



# split the dataset in train and test set
torch.manual_seed(1)
indices = torch.randperm(len(dataset)).tolist()

# train test split

test_split = 0.2
tsize = int(len(dataset)*test_split)
dataset = torch.utils.data.Subset(dataset, indices[:-tsize])
dataset_test = torch.utils.data.Subset(dataset_test, indices[-tsize:])

# define training and validation data loaders
data_loader = torch.utils.data.DataLoader(
    dataset, batch_size=5, shuffle=True, num_workers=4,
    collate_fn=utils.collate_fn)

data_loader_test = torch.utils.data.DataLoader(
    dataset_test, batch_size=5, shuffle=False, num_workers=4,
    collate_fn=utils.collate_fn)

# Model setup
def get_object_detection_model(num_classes):
    model = torchvision.models.detection.fasterrcnn_resnet50_fpn(pretrained=True)
    in_features = model.roi_heads.box_predictor.cls_score.in_features
    model.roi_heads.box_predictor = FastRCNNPredictor(in_features, num_classes)
    return model

# to train on gpu if selected.
device = torch.device('cuda') if torch.cuda.is_available() else torch.device('cpu')


num_classes = 11

# get the model using our helper function
model = get_object_detection_model(num_classes)

# move model to the right device
model.to(device)

# construct an optimizer
params = [p for p in model.parameters() if p.requires_grad]
optimizer = torch.optim.SGD(params, lr=0.005,
                            momentum=0.9, weight_decay=0.0005)

# and a learning rate scheduler which decreases the learning rate by
# 10x every 3 epochs
lr_scheduler = torch.optim.lr_scheduler.StepLR(optimizer,
                                               step_size=3,
                                               gamma=0.1)

Downloading: "https://download.pytorch.org/models/fasterrcnn_resnet50_fpn_coco-258fb6c6.pth" to /root/.cache/torch/hub/checkpoints/fasterrcnn_resnet50_fpn_coco-258fb6c6.pth
100%|██████████| 160M/160M [00:01<00:00, 85.6MB/s]

device

device(type='cuda')

# training for 8 epochs # sgd
num_epochs = 15

for epoch in range(num_epochs):
    # training for one epoch
    train_one_epoch(model, optimizer, data_loader, device, epoch, print_freq=10)
    # update the learning rate
    lr_scheduler.step()
    # evaluate on the test dataset
    evaluate(model, data_loader_test, device=device)

Epoch: [0]  [  0/104]  eta: 0:21:39  lr: 0.000053  loss: 2.5010 (2.5010)  loss_classifier: 2.0748 (2.0748)  loss_box_reg: 0.2147 (0.2147)  loss_objectness: 0.1990 (0.1990)  loss_rpn_box_reg: 0.0125 (0.0125)  time: 12.4959  data: 6.5333  max mem: 3715
Epoch: [0]  [ 10/104]  eta: 0:03:02  lr: 0.000538  loss: 2.1539 (1.9279)  loss_classifier: 1.6329 (1.4351)  loss_box_reg: 0.2379 (0.2459)  loss_objectness: 0.1990 (0.2296)  loss_rpn_box_reg: 0.0125 (0.0173)  time: 1.9377  data: 0.6165  max mem: 3875
Epoch: [0]  [ 20/104]  eta: 0:02:00  lr: 0.001023  loss: 1.2824 (1.5006)  loss_classifier: 0.7298 (1.0260)  loss_box_reg: 0.2868 (0.2820)  loss_objectness: 0.1409 (0.1747)  loss_rpn_box_reg: 0.0126 (0.0178)  time: 0.8862  data: 0.0234  max mem: 3875
Epoch: [0]  [ 30/104]  eta: 0:01:33  lr: 0.001508  loss: 1.1755 (1.3630)  loss_classifier: 0.6021 (0.8697)  loss_box_reg: 0.3267 (0.3174)  loss_objectness: 0.0952 (0.1566)  loss_rpn_box_reg: 0.0171 (0.0194)  time: 0.8888  data: 0.0215  max mem: 3875
Epoch: [0]  [ 40/104]  eta: 0:01:15  lr: 0.001993  loss: 0.9232 (1.2464)  loss_classifier: 0.4607 (0.7632)  loss_box_reg: 0.3912 (0.3269)  loss_objectness: 0.0787 (0.1373)  loss_rpn_box_reg: 0.0166 (0.0191)  time: 0.8997  data: 0.0242  max mem: 3875
Epoch: [0]  [ 50/104]  eta: 0:01:00  lr: 0.002478  loss: 0.9030 (1.1955)  loss_classifier: 0.4386 (0.7089)  loss_box_reg: 0.3942 (0.3505)  loss_objectness: 0.0389 (0.1173)  loss_rpn_box_reg: 0.0166 (0.0187)  time: 0.9106  data: 0.0249  max mem: 3875
Epoch: [0]  [ 60/104]  eta: 0:00:47  lr: 0.002963  loss: 0.9030 (1.1436)  loss_classifier: 0.4437 (0.6605)  loss_box_reg: 0.4301 (0.3621)  loss_objectness: 0.0233 (0.1029)  loss_rpn_box_reg: 0.0144 (0.0181)  time: 0.9096  data: 0.0218  max mem: 3875
Epoch: [0]  [ 70/104]  eta: 0:00:36  lr: 0.003448  loss: 0.8739 (1.0972)  loss_classifier: 0.4131 (0.6198)  loss_box_reg: 0.4052 (0.3660)  loss_objectness: 0.0335 (0.0940)  loss_rpn_box_reg: 0.0122 (0.0173)  time: 0.9186  data: 0.0233  max mem: 3875
Epoch: [0]  [ 80/104]  eta: 0:00:25  lr: 0.003933  loss: 0.7396 (1.0502)  loss_classifier: 0.3503 (0.5853)  loss_box_reg: 0.3653 (0.3625)  loss_objectness: 0.0278 (0.0857)  loss_rpn_box_reg: 0.0118 (0.0168)  time: 0.9228  data: 0.0226  max mem: 3875
Epoch: [0]  [ 90/104]  eta: 0:00:14  lr: 0.004418  loss: 0.7196 (1.0105)  loss_classifier: 0.3234 (0.5543)  loss_box_reg: 0.3427 (0.3610)  loss_objectness: 0.0169 (0.0786)  loss_rpn_box_reg: 0.0118 (0.0166)  time: 0.9814  data: 0.0218  max mem: 3875
Epoch: [0]  [100/104]  eta: 0:00:04  lr: 0.004903  loss: 0.6679 (0.9750)  loss_classifier: 0.2767 (0.5279)  loss_box_reg: 0.3131 (0.3563)  loss_objectness: 0.0202 (0.0742)  loss_rpn_box_reg: 0.0158 (0.0167)  time: 0.9938  data: 0.0228  max mem: 3875
Epoch: [0]  [103/104]  eta: 0:00:01  lr: 0.005000  loss: 0.6873 (0.9721)  loss_classifier: 0.2767 (0.5227)  loss_box_reg: 0.3400 (0.3598)  loss_objectness: 0.0206 (0.0729)  loss_rpn_box_reg: 0.0162 (0.0167)  time: 0.9364  data: 0.0207  max mem: 3875
Epoch: [0] Total time: 0:01:47 (1.0348 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:34  model_time: 0.4980 (0.4980)  evaluator_time: 0.0921 (0.0921)  time: 1.3389  data: 0.7408  max mem: 3875
Test:  [25/26]  eta: 0:00:00  model_time: 0.3987 (0.4032)  evaluator_time: 0.0276 (0.0543)  time: 0.4832  data: 0.0204  max mem: 3875
Test: Total time: 0:00:13 (0.5164 s / it)
Averaged stats: model_time: 0.3987 (0.4032)  evaluator_time: 0.0276 (0.0543)
Accumulating evaluation results...
DONE (t=0.20s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.298
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.648
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.216
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.291
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.284
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.396
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.146
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.359
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.426
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.478
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.405
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.515
Epoch: [1]  [  0/104]  eta: 0:02:56  lr: 0.005000  loss: 0.9491 (0.9491)  loss_classifier: 0.4312 (0.4312)  loss_box_reg: 0.4331 (0.4331)  loss_objectness: 0.0605 (0.0605)  loss_rpn_box_reg: 0.0243 (0.0243)  time: 1.6972  data: 0.7045  max mem: 3875
Epoch: [1]  [ 10/104]  eta: 0:01:35  lr: 0.005000  loss: 0.6119 (0.5918)  loss_classifier: 0.2030 (0.2291)  loss_box_reg: 0.3582 (0.3184)  loss_objectness: 0.0207 (0.0286)  loss_rpn_box_reg: 0.0142 (0.0157)  time: 1.0185  data: 0.0819  max mem: 3875
Epoch: [1]  [ 20/104]  eta: 0:01:22  lr: 0.005000  loss: 0.5908 (0.5789)  loss_classifier: 0.2030 (0.2234)  loss_box_reg: 0.3284 (0.3183)  loss_objectness: 0.0156 (0.0229)  loss_rpn_box_reg: 0.0127 (0.0143)  time: 0.9470  data: 0.0199  max mem: 3875
Epoch: [1]  [ 30/104]  eta: 0:01:12  lr: 0.005000  loss: 0.5204 (0.5559)  loss_classifier: 0.1771 (0.2077)  loss_box_reg: 0.3169 (0.3159)  loss_objectness: 0.0130 (0.0195)  loss_rpn_box_reg: 0.0101 (0.0127)  time: 0.9553  data: 0.0223  max mem: 3875
Epoch: [1]  [ 40/104]  eta: 0:01:02  lr: 0.005000  loss: 0.4952 (0.5393)  loss_classifier: 0.1601 (0.1970)  loss_box_reg: 0.2924 (0.3125)  loss_objectness: 0.0105 (0.0175)  loss_rpn_box_reg: 0.0084 (0.0124)  time: 0.9706  data: 0.0242  max mem: 3875
Epoch: [1]  [ 50/104]  eta: 0:00:52  lr: 0.005000  loss: 0.4952 (0.5291)  loss_classifier: 0.1601 (0.1904)  loss_box_reg: 0.3116 (0.3071)  loss_objectness: 0.0105 (0.0193)  loss_rpn_box_reg: 0.0091 (0.0123)  time: 0.9722  data: 0.0228  max mem: 3875
Epoch: [1]  [ 60/104]  eta: 0:00:43  lr: 0.005000  loss: 0.4953 (0.5243)  loss_classifier: 0.1667 (0.1868)  loss_box_reg: 0.3224 (0.3079)  loss_objectness: 0.0101 (0.0176)  loss_rpn_box_reg: 0.0093 (0.0120)  time: 0.9808  data: 0.0234  max mem: 3875
Epoch: [1]  [ 70/104]  eta: 0:00:33  lr: 0.005000  loss: 0.5289 (0.5274)  loss_classifier: 0.1667 (0.1846)  loss_box_reg: 0.3381 (0.3131)  loss_objectness: 0.0101 (0.0177)  loss_rpn_box_reg: 0.0099 (0.0120)  time: 0.9979  data: 0.0247  max mem: 3875
Epoch: [1]  [ 80/104]  eta: 0:00:23  lr: 0.005000  loss: 0.5120 (0.5213)  loss_classifier: 0.1708 (0.1816)  loss_box_reg: 0.3021 (0.3107)  loss_objectness: 0.0116 (0.0169)  loss_rpn_box_reg: 0.0117 (0.0121)  time: 1.0052  data: 0.0230  max mem: 3875
Epoch: [1]  [ 90/104]  eta: 0:00:13  lr: 0.005000  loss: 0.5059 (0.5155)  loss_classifier: 0.1356 (0.1771)  loss_box_reg: 0.2882 (0.3094)  loss_objectness: 0.0108 (0.0162)  loss_rpn_box_reg: 0.0123 (0.0128)  time: 1.0054  data: 0.0231  max mem: 3875
Epoch: [1]  [100/104]  eta: 0:00:03  lr: 0.005000  loss: 0.4005 (0.5019)  loss_classifier: 0.1111 (0.1701)  loss_box_reg: 0.2583 (0.3029)  loss_objectness: 0.0071 (0.0164)  loss_rpn_box_reg: 0.0104 (0.0125)  time: 1.0041  data: 0.0235  max mem: 3875
Epoch: [1]  [103/104]  eta: 0:00:00  lr: 0.005000  loss: 0.3574 (0.5018)  loss_classifier: 0.1061 (0.1696)  loss_box_reg: 0.2464 (0.3035)  loss_objectness: 0.0071 (0.0163)  loss_rpn_box_reg: 0.0104 (0.0125)  time: 0.9997  data: 0.0216  max mem: 3875
Epoch: [1] Total time: 0:01:42 (0.9901 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:39  model_time: 0.4493 (0.4493)  evaluator_time: 0.0767 (0.0767)  time: 1.5286  data: 0.9966  max mem: 3875
Test:  [25/26]  eta: 0:00:00  model_time: 0.4149 (0.4171)  evaluator_time: 0.0226 (0.0432)  time: 0.4765  data: 0.0200  max mem: 3875
Test: Total time: 0:00:13 (0.5298 s / it)
Averaged stats: model_time: 0.4149 (0.4171)  evaluator_time: 0.0226 (0.0432)
Accumulating evaluation results...
DONE (t=0.16s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.404
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.851
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.300
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.373
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.443
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.410
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.173
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.444
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.511
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.442
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.524
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.518
Epoch: [2]  [  0/104]  eta: 0:03:16  lr: 0.005000  loss: 0.2529 (0.2529)  loss_classifier: 0.0738 (0.0738)  loss_box_reg: 0.1640 (0.1640)  loss_objectness: 0.0093 (0.0093)  loss_rpn_box_reg: 0.0059 (0.0059)  time: 1.8854  data: 0.8509  max mem: 3875
Epoch: [2]  [ 10/104]  eta: 0:01:42  lr: 0.005000  loss: 0.3099 (0.3755)  loss_classifier: 0.0912 (0.1106)  loss_box_reg: 0.1996 (0.2446)  loss_objectness: 0.0090 (0.0085)  loss_rpn_box_reg: 0.0101 (0.0118)  time: 1.0955  data: 0.0964  max mem: 3875
Epoch: [2]  [ 20/104]  eta: 0:01:29  lr: 0.005000  loss: 0.3128 (0.3552)  loss_classifier: 0.0937 (0.1054)  loss_box_reg: 0.2095 (0.2321)  loss_objectness: 0.0053 (0.0076)  loss_rpn_box_reg: 0.0081 (0.0101)  time: 1.0247  data: 0.0217  max mem: 3875
Epoch: [2]  [ 30/104]  eta: 0:01:17  lr: 0.005000  loss: 0.3433 (0.3588)  loss_classifier: 0.0937 (0.1049)  loss_box_reg: 0.2423 (0.2373)  loss_objectness: 0.0041 (0.0068)  loss_rpn_box_reg: 0.0075 (0.0098)  time: 1.0278  data: 0.0210  max mem: 3875
Epoch: [2]  [ 40/104]  eta: 0:01:06  lr: 0.005000  loss: 0.3896 (0.3662)  loss_classifier: 0.0940 (0.1049)  loss_box_reg: 0.2579 (0.2449)  loss_objectness: 0.0050 (0.0066)  loss_rpn_box_reg: 0.0089 (0.0098)  time: 1.0151  data: 0.0201  max mem: 3875
Epoch: [2]  [ 50/104]  eta: 0:00:55  lr: 0.005000  loss: 0.3336 (0.3485)  loss_classifier: 0.0760 (0.0991)  loss_box_reg: 0.2311 (0.2342)  loss_objectness: 0.0036 (0.0061)  loss_rpn_box_reg: 0.0072 (0.0092)  time: 0.9994  data: 0.0212  max mem: 3875
Epoch: [2]  [ 60/104]  eta: 0:00:45  lr: 0.005000  loss: 0.2878 (0.3424)  loss_classifier: 0.0715 (0.0962)  loss_box_reg: 0.1954 (0.2317)  loss_objectness: 0.0022 (0.0056)  loss_rpn_box_reg: 0.0061 (0.0088)  time: 0.9889  data: 0.0225  max mem: 3875
Epoch: [2]  [ 70/104]  eta: 0:00:34  lr: 0.005000  loss: 0.3079 (0.3472)  loss_classifier: 0.0890 (0.0980)  loss_box_reg: 0.2112 (0.2347)  loss_objectness: 0.0021 (0.0058)  loss_rpn_box_reg: 0.0065 (0.0087)  time: 0.9809  data: 0.0223  max mem: 3875
Epoch: [2]  [ 80/104]  eta: 0:00:24  lr: 0.005000  loss: 0.3821 (0.3515)  loss_classifier: 0.1129 (0.0994)  loss_box_reg: 0.2273 (0.2376)  loss_objectness: 0.0038 (0.0056)  loss_rpn_box_reg: 0.0073 (0.0088)  time: 0.9853  data: 0.0229  max mem: 3875
Epoch: [2]  [ 90/104]  eta: 0:00:14  lr: 0.005000  loss: 0.4020 (0.3559)  loss_classifier: 0.1104 (0.0998)  loss_box_reg: 0.2543 (0.2404)  loss_objectness: 0.0042 (0.0059)  loss_rpn_box_reg: 0.0096 (0.0097)  time: 0.9994  data: 0.0262  max mem: 3875
Epoch: [2]  [100/104]  eta: 0:00:04  lr: 0.005000  loss: 0.3498 (0.3545)  loss_classifier: 0.0948 (0.1001)  loss_box_reg: 0.2290 (0.2388)  loss_objectness: 0.0056 (0.0059)  loss_rpn_box_reg: 0.0096 (0.0098)  time: 1.0009  data: 0.0244  max mem: 3875
Epoch: [2]  [103/104]  eta: 0:00:01  lr: 0.005000  loss: 0.3488 (0.3544)  loss_classifier: 0.0948 (0.1005)  loss_box_reg: 0.2203 (0.2382)  loss_objectness: 0.0059 (0.0059)  loss_rpn_box_reg: 0.0087 (0.0098)  time: 1.0036  data: 0.0246  max mem: 3875
Epoch: [2] Total time: 0:01:45 (1.0130 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:35  model_time: 0.5247 (0.5247)  evaluator_time: 0.0676 (0.0676)  time: 1.3611  data: 0.7346  max mem: 3875
Test:  [25/26]  eta: 0:00:00  model_time: 0.4185 (0.4221)  evaluator_time: 0.0190 (0.0292)  time: 0.4727  data: 0.0202  max mem: 3875
Test: Total time: 0:00:13 (0.5123 s / it)
Averaged stats: model_time: 0.4185 (0.4221)  evaluator_time: 0.0190 (0.0292)
Accumulating evaluation results...
DONE (t=0.21s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.427
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.879
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.365
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.407
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.443
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.493
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.196
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.471
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.528
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.490
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.513
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.582
Epoch: [3]  [  0/104]  eta: 0:03:14  lr: 0.000500  loss: 0.3512 (0.3512)  loss_classifier: 0.0984 (0.0984)  loss_box_reg: 0.2405 (0.2405)  loss_objectness: 0.0026 (0.0026)  loss_rpn_box_reg: 0.0097 (0.0097)  time: 1.8720  data: 0.8414  max mem: 3875
Epoch: [3]  [ 10/104]  eta: 0:01:43  lr: 0.000500  loss: 0.3512 (0.3440)  loss_classifier: 0.0903 (0.0885)  loss_box_reg: 0.2405 (0.2414)  loss_objectness: 0.0040 (0.0053)  loss_rpn_box_reg: 0.0098 (0.0089)  time: 1.1019  data: 0.0929  max mem: 3875
Epoch: [3]  [ 20/104]  eta: 0:01:30  lr: 0.000500  loss: 0.3133 (0.3273)  loss_classifier: 0.0721 (0.0844)  loss_box_reg: 0.2179 (0.2293)  loss_objectness: 0.0032 (0.0041)  loss_rpn_box_reg: 0.0085 (0.0095)  time: 1.0327  data: 0.0198  max mem: 3875
Epoch: [3]  [ 30/104]  eta: 0:01:18  lr: 0.000500  loss: 0.2674 (0.3052)  loss_classifier: 0.0729 (0.0806)  loss_box_reg: 0.1863 (0.2121)  loss_objectness: 0.0025 (0.0038)  loss_rpn_box_reg: 0.0073 (0.0087)  time: 1.0332  data: 0.0230  max mem: 3875
Epoch: [3]  [ 40/104]  eta: 0:01:06  lr: 0.000500  loss: 0.2504 (0.2893)  loss_classifier: 0.0707 (0.0784)  loss_box_reg: 0.1596 (0.1995)  loss_objectness: 0.0024 (0.0034)  loss_rpn_box_reg: 0.0056 (0.0080)  time: 1.0149  data: 0.0229  max mem: 3875
Epoch: [3]  [ 50/104]  eta: 0:00:55  lr: 0.000500  loss: 0.2161 (0.2790)  loss_classifier: 0.0646 (0.0765)  loss_box_reg: 0.1435 (0.1916)  loss_objectness: 0.0024 (0.0033)  loss_rpn_box_reg: 0.0054 (0.0076)  time: 0.9912  data: 0.0209  max mem: 3875
Epoch: [3]  [ 60/104]  eta: 0:00:45  lr: 0.000500  loss: 0.2449 (0.2840)  loss_classifier: 0.0796 (0.0787)  loss_box_reg: 0.1637 (0.1942)  loss_objectness: 0.0030 (0.0034)  loss_rpn_box_reg: 0.0060 (0.0077)  time: 0.9837  data: 0.0226  max mem: 3875
Epoch: [3]  [ 70/104]  eta: 0:00:34  lr: 0.000500  loss: 0.2423 (0.2732)  loss_classifier: 0.0679 (0.0758)  loss_box_reg: 0.1603 (0.1866)  loss_objectness: 0.0030 (0.0033)  loss_rpn_box_reg: 0.0055 (0.0075)  time: 0.9868  data: 0.0245  max mem: 3875
Epoch: [3]  [ 80/104]  eta: 0:00:24  lr: 0.000500  loss: 0.2168 (0.2685)  loss_classifier: 0.0586 (0.0745)  loss_box_reg: 0.1522 (0.1834)  loss_objectness: 0.0022 (0.0032)  loss_rpn_box_reg: 0.0044 (0.0074)  time: 0.9822  data: 0.0226  max mem: 3875
Epoch: [3]  [ 90/104]  eta: 0:00:14  lr: 0.000500  loss: 0.2075 (0.2656)  loss_classifier: 0.0661 (0.0737)  loss_box_reg: 0.1524 (0.1815)  loss_objectness: 0.0014 (0.0031)  loss_rpn_box_reg: 0.0058 (0.0073)  time: 0.9883  data: 0.0218  max mem: 3875
Epoch: [3]  [100/104]  eta: 0:00:04  lr: 0.000500  loss: 0.2281 (0.2643)  loss_classifier: 0.0661 (0.0741)  loss_box_reg: 0.1553 (0.1799)  loss_objectness: 0.0020 (0.0030)  loss_rpn_box_reg: 0.0062 (0.0073)  time: 1.0008  data: 0.0222  max mem: 3875
Epoch: [3]  [103/104]  eta: 0:00:01  lr: 0.000500  loss: 0.2375 (0.2639)  loss_classifier: 0.0669 (0.0740)  loss_box_reg: 0.1607 (0.1795)  loss_objectness: 0.0022 (0.0030)  loss_rpn_box_reg: 0.0062 (0.0073)  time: 0.9988  data: 0.0212  max mem: 3875
Epoch: [3] Total time: 0:01:45 (1.0120 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:38  model_time: 0.4865 (0.4865)  evaluator_time: 0.0672 (0.0672)  time: 1.4759  data: 0.9003  max mem: 3875
Test:  [25/26]  eta: 0:00:00  model_time: 0.4198 (0.4218)  evaluator_time: 0.0131 (0.0320)  time: 0.4808  data: 0.0219  max mem: 3875
Test: Total time: 0:00:13 (0.5194 s / it)
Averaged stats: model_time: 0.4198 (0.4218)  evaluator_time: 0.0131 (0.0320)
Accumulating evaluation results...
DONE (t=0.09s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.534
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.915
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.552
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.488
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.530
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.595
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.229
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.537
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.614
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.556
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.596
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.663
Epoch: [4]  [  0/104]  eta: 0:03:04  lr: 0.000500  loss: 0.2486 (0.2486)  loss_classifier: 0.0577 (0.0577)  loss_box_reg: 0.1808 (0.1808)  loss_objectness: 0.0028 (0.0028)  loss_rpn_box_reg: 0.0073 (0.0073)  time: 1.7771  data: 0.7311  max mem: 3875
Epoch: [4]  [ 10/104]  eta: 0:01:43  lr: 0.000500  loss: 0.2155 (0.2058)  loss_classifier: 0.0577 (0.0601)  loss_box_reg: 0.1410 (0.1385)  loss_objectness: 0.0026 (0.0025)  loss_rpn_box_reg: 0.0030 (0.0047)  time: 1.0965  data: 0.0827  max mem: 3875
Epoch: [4]  [ 20/104]  eta: 0:01:29  lr: 0.000500  loss: 0.1932 (0.2172)  loss_classifier: 0.0612 (0.0639)  loss_box_reg: 0.1371 (0.1459)  loss_objectness: 0.0018 (0.0024)  loss_rpn_box_reg: 0.0040 (0.0050)  time: 1.0333  data: 0.0197  max mem: 3875
Epoch: [4]  [ 30/104]  eta: 0:01:17  lr: 0.000500  loss: 0.2339 (0.2230)  loss_classifier: 0.0727 (0.0653)  loss_box_reg: 0.1525 (0.1502)  loss_objectness: 0.0017 (0.0021)  loss_rpn_box_reg: 0.0047 (0.0054)  time: 1.0278  data: 0.0210  max mem: 3875
Epoch: [4]  [ 40/104]  eta: 0:01:06  lr: 0.000500  loss: 0.2130 (0.2196)  loss_classifier: 0.0633 (0.0639)  loss_box_reg: 0.1423 (0.1482)  loss_objectness: 0.0013 (0.0021)  loss_rpn_box_reg: 0.0050 (0.0054)  time: 1.0089  data: 0.0217  max mem: 3875
Epoch: [4]  [ 50/104]  eta: 0:00:55  lr: 0.000500  loss: 0.2210 (0.2269)  loss_classifier: 0.0633 (0.0647)  loss_box_reg: 0.1510 (0.1542)  loss_objectness: 0.0013 (0.0021)  loss_rpn_box_reg: 0.0058 (0.0060)  time: 0.9999  data: 0.0236  max mem: 3875
Epoch: [4]  [ 60/104]  eta: 0:00:44  lr: 0.000500  loss: 0.2511 (0.2310)  loss_classifier: 0.0637 (0.0654)  loss_box_reg: 0.1642 (0.1573)  loss_objectness: 0.0013 (0.0021)  loss_rpn_box_reg: 0.0073 (0.0062)  time: 0.9870  data: 0.0227  max mem: 3875
Epoch: [4]  [ 70/104]  eta: 0:00:34  lr: 0.000500  loss: 0.2360 (0.2319)  loss_classifier: 0.0629 (0.0651)  loss_box_reg: 0.1642 (0.1585)  loss_objectness: 0.0014 (0.0020)  loss_rpn_box_reg: 0.0065 (0.0062)  time: 0.9781  data: 0.0222  max mem: 3875
Epoch: [4]  [ 80/104]  eta: 0:00:24  lr: 0.000500  loss: 0.2556 (0.2351)  loss_classifier: 0.0711 (0.0661)  loss_box_reg: 0.1797 (0.1603)  loss_objectness: 0.0022 (0.0023)  loss_rpn_box_reg: 0.0065 (0.0064)  time: 0.9863  data: 0.0230  max mem: 3875
Epoch: [4]  [ 90/104]  eta: 0:00:14  lr: 0.000500  loss: 0.2548 (0.2330)  loss_classifier: 0.0719 (0.0656)  loss_box_reg: 0.1665 (0.1588)  loss_objectness: 0.0020 (0.0022)  loss_rpn_box_reg: 0.0059 (0.0063)  time: 0.9923  data: 0.0226  max mem: 3875
Epoch: [4]  [100/104]  eta: 0:00:04  lr: 0.000500  loss: 0.2291 (0.2333)  loss_classifier: 0.0654 (0.0661)  loss_box_reg: 0.1484 (0.1588)  loss_objectness: 0.0011 (0.0021)  loss_rpn_box_reg: 0.0048 (0.0063)  time: 0.9973  data: 0.0218  max mem: 3875
Epoch: [4]  [103/104]  eta: 0:00:01  lr: 0.000500  loss: 0.2299 (0.2335)  loss_classifier: 0.0654 (0.0661)  loss_box_reg: 0.1570 (0.1588)  loss_objectness: 0.0011 (0.0022)  loss_rpn_box_reg: 0.0042 (0.0065)  time: 0.9997  data: 0.0221  max mem: 3875
Epoch: [4] Total time: 0:01:45 (1.0115 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:56  model_time: 0.4848 (0.4848)  evaluator_time: 0.1848 (0.1848)  time: 2.1756  data: 1.4844  max mem: 3875
Test:  [25/26]  eta: 0:00:00  model_time: 0.4164 (0.4193)  evaluator_time: 0.0129 (0.0284)  time: 0.4625  data: 0.0184  max mem: 3875
Test: Total time: 0:00:13 (0.5355 s / it)
Averaged stats: model_time: 0.4164 (0.4193)  evaluator_time: 0.0129 (0.0284)
Accumulating evaluation results...
DONE (t=0.16s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.540
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.917
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.576
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.498
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.541
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.613
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.228
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.540
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.615
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.561
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.602
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.673
Epoch: [5]  [  0/104]  eta: 0:04:26  lr: 0.000500  loss: 0.3329 (0.3329)  loss_classifier: 0.0899 (0.0899)  loss_box_reg: 0.2281 (0.2281)  loss_objectness: 0.0012 (0.0012)  loss_rpn_box_reg: 0.0139 (0.0139)  time: 2.5591  data: 1.2204  max mem: 3875
Epoch: [5]  [ 10/104]  eta: 0:01:49  lr: 0.000500  loss: 0.2360 (0.2425)  loss_classifier: 0.0729 (0.0686)  loss_box_reg: 0.1591 (0.1651)  loss_objectness: 0.0020 (0.0025)  loss_rpn_box_reg: 0.0043 (0.0063)  time: 1.1659  data: 0.1310  max mem: 3875
Epoch: [5]  [ 20/104]  eta: 0:01:32  lr: 0.000500  loss: 0.2198 (0.2314)  loss_classifier: 0.0586 (0.0661)  loss_box_reg: 0.1513 (0.1563)  loss_objectness: 0.0018 (0.0025)  loss_rpn_box_reg: 0.0047 (0.0065)  time: 1.0331  data: 0.0218  max mem: 3875
Epoch: [5]  [ 30/104]  eta: 0:01:20  lr: 0.000500  loss: 0.2059 (0.2168)  loss_classifier: 0.0529 (0.0617)  loss_box_reg: 0.1293 (0.1472)  loss_objectness: 0.0017 (0.0022)  loss_rpn_box_reg: 0.0047 (0.0056)  time: 1.0350  data: 0.0247  max mem: 3875
Epoch: [5]  [ 40/104]  eta: 0:01:07  lr: 0.000500  loss: 0.2017 (0.2115)  loss_classifier: 0.0542 (0.0598)  loss_box_reg: 0.1387 (0.1441)  loss_objectness: 0.0009 (0.0021)  loss_rpn_box_reg: 0.0037 (0.0055)  time: 1.0144  data: 0.0251  max mem: 3875
Epoch: [5]  [ 50/104]  eta: 0:00:56  lr: 0.000500  loss: 0.2017 (0.2092)  loss_classifier: 0.0527 (0.0587)  loss_box_reg: 0.1387 (0.1427)  loss_objectness: 0.0011 (0.0021)  loss_rpn_box_reg: 0.0047 (0.0057)  time: 0.9947  data: 0.0240  max mem: 3875
Epoch: [5]  [ 60/104]  eta: 0:00:45  lr: 0.000500  loss: 0.2068 (0.2098)  loss_classifier: 0.0559 (0.0589)  loss_box_reg: 0.1391 (0.1430)  loss_objectness: 0.0020 (0.0021)  loss_rpn_box_reg: 0.0048 (0.0057)  time: 0.9891  data: 0.0261  max mem: 3875
Epoch: [5]  [ 70/104]  eta: 0:00:35  lr: 0.000500  loss: 0.2366 (0.2142)  loss_classifier: 0.0650 (0.0605)  loss_box_reg: 0.1596 (0.1456)  loss_objectness: 0.0018 (0.0022)  loss_rpn_box_reg: 0.0058 (0.0060)  time: 0.9862  data: 0.0246  max mem: 3875
Epoch: [5]  [ 80/104]  eta: 0:00:24  lr: 0.000500  loss: 0.2242 (0.2169)  loss_classifier: 0.0663 (0.0612)  loss_box_reg: 0.1553 (0.1477)  loss_objectness: 0.0015 (0.0021)  loss_rpn_box_reg: 0.0058 (0.0059)  time: 0.9817  data: 0.0216  max mem: 3875
Epoch: [5]  [ 90/104]  eta: 0:00:14  lr: 0.000500  loss: 0.2099 (0.2160)  loss_classifier: 0.0635 (0.0611)  loss_box_reg: 0.1410 (0.1470)  loss_objectness: 0.0010 (0.0021)  loss_rpn_box_reg: 0.0046 (0.0059)  time: 0.9899  data: 0.0235  max mem: 3875
Epoch: [5]  [100/104]  eta: 0:00:04  lr: 0.000500  loss: 0.2310 (0.2225)  loss_classifier: 0.0614 (0.0622)  loss_box_reg: 0.1553 (0.1520)  loss_objectness: 0.0022 (0.0023)  loss_rpn_box_reg: 0.0050 (0.0061)  time: 1.0036  data: 0.0247  max mem: 3875
Epoch: [5]  [103/104]  eta: 0:00:01  lr: 0.000500  loss: 0.2390 (0.2231)  loss_classifier: 0.0644 (0.0622)  loss_box_reg: 0.1553 (0.1524)  loss_objectness: 0.0025 (0.0023)  loss_rpn_box_reg: 0.0050 (0.0062)  time: 1.0010  data: 0.0226  max mem: 3875
Epoch: [5] Total time: 0:01:46 (1.0203 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:53  model_time: 0.4563 (0.4563)  evaluator_time: 0.0650 (0.0650)  time: 2.0548  data: 1.5238  max mem: 3875
Test:  [25/26]  eta: 0:00:00  model_time: 0.4182 (0.4190)  evaluator_time: 0.0120 (0.0300)  time: 0.4740  data: 0.0207  max mem: 3875
Test: Total time: 0:00:13 (0.5384 s / it)
Averaged stats: model_time: 0.4182 (0.4190)  evaluator_time: 0.0120 (0.0300)
Accumulating evaluation results...
DONE (t=0.08s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.543
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.928
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.561
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.477
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.544
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.635
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.234
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.544
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.610
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.541
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.597
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.678
Epoch: [6]  [  0/104]  eta: 0:03:10  lr: 0.000050  loss: 0.1748 (0.1748)  loss_classifier: 0.0497 (0.0497)  loss_box_reg: 0.1187 (0.1187)  loss_objectness: 0.0005 (0.0005)  loss_rpn_box_reg: 0.0059 (0.0059)  time: 1.8283  data: 0.7444  max mem: 3875
Epoch: [6]  [ 10/104]  eta: 0:01:43  lr: 0.000050  loss: 0.2450 (0.2322)  loss_classifier: 0.0689 (0.0688)  loss_box_reg: 0.1692 (0.1551)  loss_objectness: 0.0016 (0.0019)  loss_rpn_box_reg: 0.0063 (0.0064)  time: 1.0963  data: 0.0858  max mem: 3875
Epoch: [6]  [ 20/104]  eta: 0:01:29  lr: 0.000050  loss: 0.2194 (0.2229)  loss_classifier: 0.0646 (0.0635)  loss_box_reg: 0.1513 (0.1514)  loss_objectness: 0.0016 (0.0021)  loss_rpn_box_reg: 0.0063 (0.0060)  time: 1.0323  data: 0.0208  max mem: 3875
Epoch: [6]  [ 30/104]  eta: 0:01:17  lr: 0.000050  loss: 0.2048 (0.2163)  loss_classifier: 0.0552 (0.0611)  loss_box_reg: 0.1435 (0.1477)  loss_objectness: 0.0011 (0.0020)  loss_rpn_box_reg: 0.0041 (0.0055)  time: 1.0301  data: 0.0215  max mem: 3875
Epoch: [6]  [ 40/104]  eta: 0:01:06  lr: 0.000050  loss: 0.1756 (0.2070)  loss_classifier: 0.0488 (0.0574)  loss_box_reg: 0.1240 (0.1425)  loss_objectness: 0.0011 (0.0018)  loss_rpn_box_reg: 0.0042 (0.0054)  time: 1.0108  data: 0.0221  max mem: 3875
Epoch: [6]  [ 50/104]  eta: 0:00:55  lr: 0.000050  loss: 0.1981 (0.2085)  loss_classifier: 0.0522 (0.0581)  loss_box_reg: 0.1356 (0.1429)  loss_objectness: 0.0009 (0.0019)  loss_rpn_box_reg: 0.0046 (0.0056)  time: 0.9953  data: 0.0228  max mem: 3875
Epoch: [6]  [ 60/104]  eta: 0:00:44  lr: 0.000050  loss: 0.2118 (0.2096)  loss_classifier: 0.0613 (0.0586)  loss_box_reg: 0.1436 (0.1435)  loss_objectness: 0.0013 (0.0019)  loss_rpn_box_reg: 0.0044 (0.0057)  time: 0.9844  data: 0.0232  max mem: 3875
Epoch: [6]  [ 70/104]  eta: 0:00:34  lr: 0.000050  loss: 0.2118 (0.2107)  loss_classifier: 0.0564 (0.0588)  loss_box_reg: 0.1498 (0.1442)  loss_objectness: 0.0008 (0.0019)  loss_rpn_box_reg: 0.0053 (0.0057)  time: 0.9816  data: 0.0236  max mem: 3875
Epoch: [6]  [ 80/104]  eta: 0:00:24  lr: 0.000050  loss: 0.2146 (0.2110)  loss_classifier: 0.0564 (0.0587)  loss_box_reg: 0.1474 (0.1444)  loss_objectness: 0.0009 (0.0019)  loss_rpn_box_reg: 0.0053 (0.0059)  time: 0.9904  data: 0.0260  max mem: 3875
Epoch: [6]  [ 90/104]  eta: 0:00:14  lr: 0.000050  loss: 0.2061 (0.2114)  loss_classifier: 0.0595 (0.0589)  loss_box_reg: 0.1447 (0.1446)  loss_objectness: 0.0014 (0.0019)  loss_rpn_box_reg: 0.0050 (0.0059)  time: 0.9995  data: 0.0258  max mem: 3875
Epoch: [6]  [100/104]  eta: 0:00:04  lr: 0.000050  loss: 0.2095 (0.2115)  loss_classifier: 0.0598 (0.0592)  loss_box_reg: 0.1422 (0.1445)  loss_objectness: 0.0012 (0.0019)  loss_rpn_box_reg: 0.0050 (0.0059)  time: 0.9972  data: 0.0219  max mem: 3875
Epoch: [6]  [103/104]  eta: 0:00:01  lr: 0.000050  loss: 0.2186 (0.2117)  loss_classifier: 0.0608 (0.0593)  loss_box_reg: 0.1447 (0.1446)  loss_objectness: 0.0013 (0.0019)  loss_rpn_box_reg: 0.0050 (0.0059)  time: 1.0005  data: 0.0220  max mem: 3875
Epoch: [6] Total time: 0:01:45 (1.0129 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:38  model_time: 0.4612 (0.4612)  evaluator_time: 0.0828 (0.0828)  time: 1.4617  data: 0.8888  max mem: 3875
Test:  [25/26]  eta: 0:00:00  model_time: 0.4153 (0.4177)  evaluator_time: 0.0128 (0.0239)  time: 0.4611  data: 0.0183  max mem: 3875
Test: Total time: 0:00:13 (0.5041 s / it)
Averaged stats: model_time: 0.4153 (0.4177)  evaluator_time: 0.0128 (0.0239)
Accumulating evaluation results...
DONE (t=0.08s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.548
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.928
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.573
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.482
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.549
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.640
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.236
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.544
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.613
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.546
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.599
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.685
Epoch: [7]  [  0/104]  eta: 0:03:03  lr: 0.000050  loss: 0.3377 (0.3377)  loss_classifier: 0.0893 (0.0893)  loss_box_reg: 0.2355 (0.2355)  loss_objectness: 0.0032 (0.0032)  loss_rpn_box_reg: 0.0097 (0.0097)  time: 1.7651  data: 0.7430  max mem: 3875
Epoch: [7]  [ 10/104]  eta: 0:01:42  lr: 0.000050  loss: 0.2396 (0.2352)  loss_classifier: 0.0663 (0.0656)  loss_box_reg: 0.1588 (0.1608)  loss_objectness: 0.0014 (0.0023)  loss_rpn_box_reg: 0.0060 (0.0066)  time: 1.0899  data: 0.0860  max mem: 3875
Epoch: [7]  [ 20/104]  eta: 0:01:29  lr: 0.000050  loss: 0.1979 (0.2177)  loss_classifier: 0.0567 (0.0607)  loss_box_reg: 0.1329 (0.1488)  loss_objectness: 0.0010 (0.0019)  loss_rpn_box_reg: 0.0058 (0.0063)  time: 1.0304  data: 0.0212  max mem: 3875
Epoch: [7]  [ 30/104]  eta: 0:01:17  lr: 0.000050  loss: 0.1922 (0.2192)  loss_classifier: 0.0567 (0.0634)  loss_box_reg: 0.1329 (0.1474)  loss_objectness: 0.0010 (0.0018)  loss_rpn_box_reg: 0.0049 (0.0066)  time: 1.0332  data: 0.0228  max mem: 3875
Epoch: [7]  [ 40/104]  eta: 0:01:06  lr: 0.000050  loss: 0.2160 (0.2161)  loss_classifier: 0.0579 (0.0619)  loss_box_reg: 0.1450 (0.1458)  loss_objectness: 0.0011 (0.0017)  loss_rpn_box_reg: 0.0059 (0.0066)  time: 1.0172  data: 0.0243  max mem: 3875
Epoch: [7]  [ 50/104]  eta: 0:00:55  lr: 0.000050  loss: 0.2120 (0.2142)  loss_classifier: 0.0539 (0.0604)  loss_box_reg: 0.1449 (0.1458)  loss_objectness: 0.0011 (0.0017)  loss_rpn_box_reg: 0.0053 (0.0062)  time: 0.9945  data: 0.0235  max mem: 3875
Epoch: [7]  [ 60/104]  eta: 0:00:44  lr: 0.000050  loss: 0.1899 (0.2105)  loss_classifier: 0.0537 (0.0594)  loss_box_reg: 0.1309 (0.1433)  loss_objectness: 0.0019 (0.0018)  loss_rpn_box_reg: 0.0042 (0.0060)  time: 0.9808  data: 0.0224  max mem: 3875
Epoch: [7]  [ 70/104]  eta: 0:00:34  lr: 0.000050  loss: 0.1954 (0.2118)  loss_classifier: 0.0578 (0.0595)  loss_box_reg: 0.1323 (0.1444)  loss_objectness: 0.0016 (0.0018)  loss_rpn_box_reg: 0.0052 (0.0061)  time: 0.9809  data: 0.0232  max mem: 3875
Epoch: [7]  [ 80/104]  eta: 0:00:24  lr: 0.000050  loss: 0.2007 (0.2075)  loss_classifier: 0.0570 (0.0583)  loss_box_reg: 0.1323 (0.1416)  loss_objectness: 0.0016 (0.0018)  loss_rpn_box_reg: 0.0048 (0.0058)  time: 0.9861  data: 0.0224  max mem: 3875
Epoch: [7]  [ 90/104]  eta: 0:00:14  lr: 0.000050  loss: 0.1722 (0.2061)  loss_classifier: 0.0577 (0.0582)  loss_box_reg: 0.1187 (0.1403)  loss_objectness: 0.0011 (0.0018)  loss_rpn_box_reg: 0.0042 (0.0058)  time: 0.9900  data: 0.0216  max mem: 3875
Epoch: [7]  [100/104]  eta: 0:00:04  lr: 0.000050  loss: 0.2126 (0.2080)  loss_classifier: 0.0583 (0.0585)  loss_box_reg: 0.1433 (0.1418)  loss_objectness: 0.0008 (0.0018)  loss_rpn_box_reg: 0.0052 (0.0059)  time: 0.9945  data: 0.0229  max mem: 3875
Epoch: [7]  [103/104]  eta: 0:00:01  lr: 0.000050  loss: 0.2127 (0.2091)  loss_classifier: 0.0603 (0.0586)  loss_box_reg: 0.1433 (0.1428)  loss_objectness: 0.0008 (0.0018)  loss_rpn_box_reg: 0.0050 (0.0059)  time: 0.9928  data: 0.0218  max mem: 3875
Epoch: [7] Total time: 0:01:45 (1.0097 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:38  model_time: 0.5028 (0.5028)  evaluator_time: 0.1567 (0.1567)  time: 1.4818  data: 0.8029  max mem: 3875
Test:  [25/26]  eta: 0:00:00  model_time: 0.4149 (0.4197)  evaluator_time: 0.0106 (0.0323)  time: 0.4615  data: 0.0184  max mem: 3875
Test: Total time: 0:00:13 (0.5105 s / it)
Averaged stats: model_time: 0.4149 (0.4197)  evaluator_time: 0.0106 (0.0323)
Accumulating evaluation results...
DONE (t=0.08s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.547
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.928
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.580
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.486
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.547
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.627
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.234
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.543
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.613
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.549
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.599
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.673
Epoch: [8]  [  0/104]  eta: 0:03:03  lr: 0.000050  loss: 0.1904 (0.1904)  loss_classifier: 0.0511 (0.0511)  loss_box_reg: 0.1332 (0.1332)  loss_objectness: 0.0031 (0.0031)  loss_rpn_box_reg: 0.0030 (0.0030)  time: 1.7656  data: 0.7269  max mem: 3875
Epoch: [8]  [ 10/104]  eta: 0:01:42  lr: 0.000050  loss: 0.1936 (0.2069)  loss_classifier: 0.0489 (0.0585)  loss_box_reg: 0.1332 (0.1405)  loss_objectness: 0.0021 (0.0023)  loss_rpn_box_reg: 0.0032 (0.0057)  time: 1.0956  data: 0.0855  max mem: 3875
Epoch: [8]  [ 20/104]  eta: 0:01:29  lr: 0.000050  loss: 0.1984 (0.1999)  loss_classifier: 0.0493 (0.0574)  loss_box_reg: 0.1366 (0.1350)  loss_objectness: 0.0012 (0.0021)  loss_rpn_box_reg: 0.0042 (0.0055)  time: 1.0346  data: 0.0230  max mem: 3875
Epoch: [8]  [ 30/104]  eta: 0:01:17  lr: 0.000050  loss: 0.2162 (0.2172)  loss_classifier: 0.0580 (0.0618)  loss_box_reg: 0.1442 (0.1470)  loss_objectness: 0.0018 (0.0021)  loss_rpn_box_reg: 0.0046 (0.0062)  time: 1.0295  data: 0.0229  max mem: 3875
Epoch: [8]  [ 40/104]  eta: 0:01:06  lr: 0.000050  loss: 0.2084 (0.2106)  loss_classifier: 0.0619 (0.0616)  loss_box_reg: 0.1368 (0.1412)  loss_objectness: 0.0013 (0.0020)  loss_rpn_box_reg: 0.0042 (0.0057)  time: 1.0076  data: 0.0210  max mem: 3875
Epoch: [8]  [ 50/104]  eta: 0:00:55  lr: 0.000050  loss: 0.2009 (0.2107)  loss_classifier: 0.0622 (0.0618)  loss_box_reg: 0.1270 (0.1413)  loss_objectness: 0.0009 (0.0018)  loss_rpn_box_reg: 0.0042 (0.0057)  time: 0.9915  data: 0.0215  max mem: 3875
Epoch: [8]  [ 60/104]  eta: 0:00:44  lr: 0.000050  loss: 0.2150 (0.2111)  loss_classifier: 0.0575 (0.0608)  loss_box_reg: 0.1517 (0.1426)  loss_objectness: 0.0011 (0.0018)  loss_rpn_box_reg: 0.0052 (0.0059)  time: 0.9845  data: 0.0227  max mem: 3875
Epoch: [8]  [ 70/104]  eta: 0:00:34  lr: 0.000050  loss: 0.2150 (0.2114)  loss_classifier: 0.0575 (0.0605)  loss_box_reg: 0.1531 (0.1431)  loss_objectness: 0.0019 (0.0018)  loss_rpn_box_reg: 0.0065 (0.0060)  time: 0.9791  data: 0.0220  max mem: 3875
Epoch: [8]  [ 80/104]  eta: 0:00:24  lr: 0.000050  loss: 0.2228 (0.2130)  loss_classifier: 0.0610 (0.0605)  loss_box_reg: 0.1561 (0.1446)  loss_objectness: 0.0020 (0.0018)  loss_rpn_box_reg: 0.0068 (0.0061)  time: 0.9812  data: 0.0217  max mem: 3875
Epoch: [8]  [ 90/104]  eta: 0:00:14  lr: 0.000050  loss: 0.2226 (0.2116)  loss_classifier: 0.0574 (0.0602)  loss_box_reg: 0.1549 (0.1436)  loss_objectness: 0.0014 (0.0018)  loss_rpn_box_reg: 0.0061 (0.0060)  time: 0.9908  data: 0.0226  max mem: 3875
Epoch: [8]  [100/104]  eta: 0:00:04  lr: 0.000050  loss: 0.1706 (0.2097)  loss_classifier: 0.0499 (0.0593)  loss_box_reg: 0.1162 (0.1427)  loss_objectness: 0.0010 (0.0018)  loss_rpn_box_reg: 0.0048 (0.0059)  time: 0.9941  data: 0.0213  max mem: 3875
Epoch: [8]  [103/104]  eta: 0:00:01  lr: 0.000050  loss: 0.1645 (0.2086)  loss_classifier: 0.0468 (0.0588)  loss_box_reg: 0.1145 (0.1422)  loss_objectness: 0.0007 (0.0017)  loss_rpn_box_reg: 0.0039 (0.0059)  time: 0.9967  data: 0.0211  max mem: 3875
Epoch: [8] Total time: 0:01:44 (1.0086 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:36  model_time: 0.4546 (0.4546)  evaluator_time: 0.0618 (0.0618)  time: 1.3852  data: 0.8516  max mem: 3875
Test:  [25/26]  eta: 0:00:00  model_time: 0.4210 (0.4196)  evaluator_time: 0.0127 (0.0233)  time: 0.4646  data: 0.0182  max mem: 3875
Test: Total time: 0:00:13 (0.5037 s / it)
Averaged stats: model_time: 0.4210 (0.4196)  evaluator_time: 0.0127 (0.0233)
Accumulating evaluation results...
DONE (t=0.08s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.548
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.928
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.587
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.486
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.546
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.627
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.235
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.543
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.614
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.547
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.601
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.673
Epoch: [9]  [  0/104]  eta: 0:03:09  lr: 0.000005  loss: 0.1976 (0.1976)  loss_classifier: 0.0576 (0.0576)  loss_box_reg: 0.1314 (0.1314)  loss_objectness: 0.0030 (0.0030)  loss_rpn_box_reg: 0.0056 (0.0056)  time: 1.8196  data: 0.7632  max mem: 3875
Epoch: [9]  [ 10/104]  eta: 0:01:43  lr: 0.000005  loss: 0.1904 (0.1900)  loss_classifier: 0.0542 (0.0547)  loss_box_reg: 0.1302 (0.1282)  loss_objectness: 0.0014 (0.0020)  loss_rpn_box_reg: 0.0046 (0.0051)  time: 1.0982  data: 0.0892  max mem: 3875
Epoch: [9]  [ 20/104]  eta: 0:01:29  lr: 0.000005  loss: 0.1882 (0.1912)  loss_classifier: 0.0529 (0.0547)  loss_box_reg: 0.1255 (0.1301)  loss_objectness: 0.0013 (0.0017)  loss_rpn_box_reg: 0.0040 (0.0048)  time: 1.0271  data: 0.0208  max mem: 3875
Epoch: [9]  [ 30/104]  eta: 0:01:17  lr: 0.000005  loss: 0.1947 (0.1902)  loss_classifier: 0.0508 (0.0538)  loss_box_reg: 0.1255 (0.1299)  loss_objectness: 0.0012 (0.0016)  loss_rpn_box_reg: 0.0040 (0.0049)  time: 1.0192  data: 0.0215  max mem: 3875
Epoch: [9]  [ 40/104]  eta: 0:01:06  lr: 0.000005  loss: 0.1974 (0.1946)  loss_classifier: 0.0534 (0.0546)  loss_box_reg: 0.1315 (0.1329)  loss_objectness: 0.0013 (0.0016)  loss_rpn_box_reg: 0.0043 (0.0054)  time: 1.0052  data: 0.0221  max mem: 3875
Epoch: [9]  [ 50/104]  eta: 0:00:55  lr: 0.000005  loss: 0.1927 (0.1957)  loss_classifier: 0.0568 (0.0554)  loss_box_reg: 0.1315 (0.1333)  loss_objectness: 0.0012 (0.0015)  loss_rpn_box_reg: 0.0052 (0.0055)  time: 0.9902  data: 0.0203  max mem: 3875
Epoch: [9]  [ 60/104]  eta: 0:00:44  lr: 0.000005  loss: 0.2107 (0.1974)  loss_classifier: 0.0577 (0.0556)  loss_box_reg: 0.1355 (0.1346)  loss_objectness: 0.0009 (0.0016)  loss_rpn_box_reg: 0.0054 (0.0057)  time: 0.9783  data: 0.0208  max mem: 3875
Epoch: [9]  [ 70/104]  eta: 0:00:34  lr: 0.000005  loss: 0.2091 (0.1986)  loss_classifier: 0.0561 (0.0563)  loss_box_reg: 0.1366 (0.1352)  loss_objectness: 0.0008 (0.0016)  loss_rpn_box_reg: 0.0043 (0.0055)  time: 0.9819  data: 0.0225  max mem: 3875
Epoch: [9]  [ 80/104]  eta: 0:00:24  lr: 0.000005  loss: 0.2091 (0.2051)  loss_classifier: 0.0641 (0.0586)  loss_box_reg: 0.1418 (0.1391)  loss_objectness: 0.0012 (0.0017)  loss_rpn_box_reg: 0.0053 (0.0058)  time: 0.9880  data: 0.0223  max mem: 3875
Epoch: [9]  [ 90/104]  eta: 0:00:14  lr: 0.000005  loss: 0.2247 (0.2080)  loss_classifier: 0.0641 (0.0592)  loss_box_reg: 0.1488 (0.1413)  loss_objectness: 0.0014 (0.0017)  loss_rpn_box_reg: 0.0055 (0.0058)  time: 0.9888  data: 0.0214  max mem: 3875
Epoch: [9]  [100/104]  eta: 0:00:04  lr: 0.000005  loss: 0.2146 (0.2070)  loss_classifier: 0.0593 (0.0586)  loss_box_reg: 0.1465 (0.1409)  loss_objectness: 0.0011 (0.0017)  loss_rpn_box_reg: 0.0054 (0.0057)  time: 0.9937  data: 0.0217  max mem: 3875
Epoch: [9]  [103/104]  eta: 0:00:01  lr: 0.000005  loss: 0.2110 (0.2080)  loss_classifier: 0.0592 (0.0591)  loss_box_reg: 0.1438 (0.1413)  loss_objectness: 0.0010 (0.0017)  loss_rpn_box_reg: 0.0053 (0.0059)  time: 0.9943  data: 0.0213  max mem: 3875
Epoch: [9] Total time: 0:01:44 (1.0075 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:54  model_time: 0.5491 (0.5491)  evaluator_time: 0.2153 (0.2153)  time: 2.0892  data: 1.3075  max mem: 3875
Test:  [25/26]  eta: 0:00:00  model_time: 0.4136 (0.4206)  evaluator_time: 0.0113 (0.0299)  time: 0.4544  data: 0.0166  max mem: 3875
Test: Total time: 0:00:13 (0.5293 s / it)
Averaged stats: model_time: 0.4136 (0.4206)  evaluator_time: 0.0113 (0.0299)
Accumulating evaluation results...
DONE (t=0.14s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.548
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.928
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.587
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.486
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.547
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.627
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.235
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.543
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.615
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.547
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.601
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.673
Epoch: [10]  [  0/104]  eta: 0:04:01  lr: 0.000005  loss: 0.2751 (0.2751)  loss_classifier: 0.0831 (0.0831)  loss_box_reg: 0.1825 (0.1825)  loss_objectness: 0.0013 (0.0013)  loss_rpn_box_reg: 0.0083 (0.0083)  time: 2.3237  data: 1.2064  max mem: 3875
Epoch: [10]  [ 10/104]  eta: 0:01:47  lr: 0.000005  loss: 0.2280 (0.2265)  loss_classifier: 0.0661 (0.0682)  loss_box_reg: 0.1609 (0.1499)  loss_objectness: 0.0017 (0.0021)  loss_rpn_box_reg: 0.0055 (0.0062)  time: 1.1399  data: 0.1263  max mem: 3875
Epoch: [10]  [ 20/104]  eta: 0:01:31  lr: 0.000005  loss: 0.1922 (0.2111)  loss_classifier: 0.0605 (0.0607)  loss_box_reg: 0.1324 (0.1427)  loss_objectness: 0.0014 (0.0019)  loss_rpn_box_reg: 0.0047 (0.0058)  time: 1.0319  data: 0.0200  max mem: 3875
Epoch: [10]  [ 30/104]  eta: 0:01:19  lr: 0.000005  loss: 0.2016 (0.2236)  loss_classifier: 0.0590 (0.0650)  loss_box_reg: 0.1326 (0.1502)  loss_objectness: 0.0014 (0.0019)  loss_rpn_box_reg: 0.0060 (0.0065)  time: 1.0328  data: 0.0218  max mem: 3875
Epoch: [10]  [ 40/104]  eta: 0:01:07  lr: 0.000005  loss: 0.1926 (0.2149)  loss_classifier: 0.0549 (0.0622)  loss_box_reg: 0.1342 (0.1445)  loss_objectness: 0.0009 (0.0018)  loss_rpn_box_reg: 0.0041 (0.0065)  time: 1.0111  data: 0.0215  max mem: 3875
Epoch: [10]  [ 50/104]  eta: 0:00:56  lr: 0.000005  loss: 0.1722 (0.2061)  loss_classifier: 0.0491 (0.0592)  loss_box_reg: 0.1142 (0.1391)  loss_objectness: 0.0006 (0.0017)  loss_rpn_box_reg: 0.0038 (0.0061)  time: 0.9925  data: 0.0218  max mem: 3875
Epoch: [10]  [ 60/104]  eta: 0:00:45  lr: 0.000005  loss: 0.1990 (0.2091)  loss_classifier: 0.0547 (0.0599)  loss_box_reg: 0.1351 (0.1412)  loss_objectness: 0.0018 (0.0018)  loss_rpn_box_reg: 0.0056 (0.0061)  time: 0.9838  data: 0.0221  max mem: 3875
Epoch: [10]  [ 70/104]  eta: 0:00:34  lr: 0.000005  loss: 0.2036 (0.2068)  loss_classifier: 0.0572 (0.0591)  loss_box_reg: 0.1417 (0.1401)  loss_objectness: 0.0015 (0.0018)  loss_rpn_box_reg: 0.0048 (0.0058)  time: 0.9812  data: 0.0233  max mem: 3875
Epoch: [10]  [ 80/104]  eta: 0:00:24  lr: 0.000005  loss: 0.2014 (0.2067)  loss_classifier: 0.0511 (0.0585)  loss_box_reg: 0.1399 (0.1405)  loss_objectness: 0.0014 (0.0019)  loss_rpn_box_reg: 0.0036 (0.0058)  time: 0.9775  data: 0.0225  max mem: 3875
Epoch: [10]  [ 90/104]  eta: 0:00:14  lr: 0.000005  loss: 0.2014 (0.2059)  loss_classifier: 0.0546 (0.0583)  loss_box_reg: 0.1412 (0.1400)  loss_objectness: 0.0015 (0.0019)  loss_rpn_box_reg: 0.0036 (0.0057)  time: 0.9832  data: 0.0213  max mem: 3875
Epoch: [10]  [100/104]  eta: 0:00:04  lr: 0.000005  loss: 0.1887 (0.2061)  loss_classifier: 0.0589 (0.0583)  loss_box_reg: 0.1298 (0.1401)  loss_objectness: 0.0015 (0.0019)  loss_rpn_box_reg: 0.0035 (0.0058)  time: 0.9930  data: 0.0208  max mem: 3875
Epoch: [10]  [103/104]  eta: 0:00:01  lr: 0.000005  loss: 0.1882 (0.2076)  loss_classifier: 0.0546 (0.0587)  loss_box_reg: 0.1339 (0.1411)  loss_objectness: 0.0016 (0.0020)  loss_rpn_box_reg: 0.0034 (0.0058)  time: 0.9903  data: 0.0196  max mem: 3875
Epoch: [10] Total time: 0:01:45 (1.0130 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:38  model_time: 0.4692 (0.4692)  evaluator_time: 0.0643 (0.0643)  time: 1.4834  data: 0.9290  max mem: 3875
Test:  [25/26]  eta: 0:00:00  model_time: 0.4167 (0.4182)  evaluator_time: 0.0112 (0.0283)  time: 0.4725  data: 0.0199  max mem: 3875
Test: Total time: 0:00:13 (0.5124 s / it)
Averaged stats: model_time: 0.4167 (0.4182)  evaluator_time: 0.0112 (0.0283)
Accumulating evaluation results...
DONE (t=0.08s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.548
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.928
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.586
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.487
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.547
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.627
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.236
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.544
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.615
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.548
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.602
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.673
Epoch: [11]  [  0/104]  eta: 0:03:03  lr: 0.000005  loss: 0.1635 (0.1635)  loss_classifier: 0.0396 (0.0396)  loss_box_reg: 0.1188 (0.1188)  loss_objectness: 0.0006 (0.0006)  loss_rpn_box_reg: 0.0046 (0.0046)  time: 1.7666  data: 0.7019  max mem: 3875
Epoch: [11]  [ 10/104]  eta: 0:01:42  lr: 0.000005  loss: 0.1990 (0.1852)  loss_classifier: 0.0505 (0.0493)  loss_box_reg: 0.1402 (0.1305)  loss_objectness: 0.0007 (0.0008)  loss_rpn_box_reg: 0.0045 (0.0046)  time: 1.0910  data: 0.0818  max mem: 3875
Epoch: [11]  [ 20/104]  eta: 0:01:29  lr: 0.000005  loss: 0.1990 (0.1972)  loss_classifier: 0.0505 (0.0548)  loss_box_reg: 0.1385 (0.1357)  loss_objectness: 0.0010 (0.0015)  loss_rpn_box_reg: 0.0049 (0.0053)  time: 1.0275  data: 0.0204  max mem: 3875
Epoch: [11]  [ 30/104]  eta: 0:01:17  lr: 0.000005  loss: 0.2375 (0.2158)  loss_classifier: 0.0670 (0.0603)  loss_box_reg: 0.1542 (0.1479)  loss_objectness: 0.0018 (0.0020)  loss_rpn_box_reg: 0.0060 (0.0056)  time: 1.0214  data: 0.0209  max mem: 3875
Epoch: [11]  [ 40/104]  eta: 0:01:06  lr: 0.000005  loss: 0.2013 (0.2082)  loss_classifier: 0.0544 (0.0581)  loss_box_reg: 0.1448 (0.1425)  loss_objectness: 0.0014 (0.0018)  loss_rpn_box_reg: 0.0052 (0.0057)  time: 1.0068  data: 0.0226  max mem: 3875
Epoch: [11]  [ 50/104]  eta: 0:00:55  lr: 0.000005  loss: 0.1670 (0.2043)  loss_classifier: 0.0510 (0.0574)  loss_box_reg: 0.1161 (0.1394)  loss_objectness: 0.0013 (0.0019)  loss_rpn_box_reg: 0.0046 (0.0056)  time: 0.9946  data: 0.0237  max mem: 3875
Epoch: [11]  [ 60/104]  eta: 0:00:44  lr: 0.000005  loss: 0.1670 (0.2026)  loss_classifier: 0.0510 (0.0570)  loss_box_reg: 0.1170 (0.1380)  loss_objectness: 0.0012 (0.0021)  loss_rpn_box_reg: 0.0043 (0.0055)  time: 0.9816  data: 0.0222  max mem: 3875
Epoch: [11]  [ 70/104]  eta: 0:00:34  lr: 0.000005  loss: 0.1836 (0.2035)  loss_classifier: 0.0537 (0.0571)  loss_box_reg: 0.1204 (0.1388)  loss_objectness: 0.0008 (0.0020)  loss_rpn_box_reg: 0.0040 (0.0056)  time: 0.9792  data: 0.0227  max mem: 3875
Epoch: [11]  [ 80/104]  eta: 0:00:24  lr: 0.000005  loss: 0.2026 (0.2034)  loss_classifier: 0.0572 (0.0570)  loss_box_reg: 0.1380 (0.1388)  loss_objectness: 0.0008 (0.0019)  loss_rpn_box_reg: 0.0053 (0.0057)  time: 0.9844  data: 0.0232  max mem: 3875
Epoch: [11]  [ 90/104]  eta: 0:00:14  lr: 0.000005  loss: 0.2026 (0.2047)  loss_classifier: 0.0541 (0.0572)  loss_box_reg: 0.1389 (0.1400)  loss_objectness: 0.0010 (0.0019)  loss_rpn_box_reg: 0.0053 (0.0057)  time: 0.9930  data: 0.0228  max mem: 3875
Epoch: [11]  [100/104]  eta: 0:00:04  lr: 0.000005  loss: 0.1901 (0.2054)  loss_classifier: 0.0549 (0.0579)  loss_box_reg: 0.1389 (0.1400)  loss_objectness: 0.0016 (0.0019)  loss_rpn_box_reg: 0.0058 (0.0057)  time: 0.9939  data: 0.0210  max mem: 3875
Epoch: [11]  [103/104]  eta: 0:00:01  lr: 0.000005  loss: 0.2141 (0.2069)  loss_classifier: 0.0549 (0.0582)  loss_box_reg: 0.1521 (0.1409)  loss_objectness: 0.0016 (0.0019)  loss_rpn_box_reg: 0.0065 (0.0058)  time: 0.9972  data: 0.0210  max mem: 3875
Epoch: [11] Total time: 0:01:44 (1.0078 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:49  model_time: 0.4933 (0.4933)  evaluator_time: 0.0623 (0.0623)  time: 1.8962  data: 1.3249  max mem: 3875
Test:  [25/26]  eta: 0:00:00  model_time: 0.4157 (0.4191)  evaluator_time: 0.0114 (0.0227)  time: 0.4753  data: 0.0326  max mem: 3875
Test: Total time: 0:00:13 (0.5335 s / it)
Averaged stats: model_time: 0.4157 (0.4191)  evaluator_time: 0.0114 (0.0227)
Accumulating evaluation results...
DONE (t=0.14s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.549
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.928
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.585
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.487
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.547
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.627
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.236
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.544
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.615
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.548
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.602
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.673
Epoch: [12]  [  0/104]  eta: 0:03:26  lr: 0.000001  loss: 0.1430 (0.1430)  loss_classifier: 0.0349 (0.0349)  loss_box_reg: 0.1038 (0.1038)  loss_objectness: 0.0007 (0.0007)  loss_rpn_box_reg: 0.0036 (0.0036)  time: 1.9831  data: 0.9518  max mem: 3875
Epoch: [12]  [ 10/104]  eta: 0:01:43  lr: 0.000001  loss: 0.2299 (0.2337)  loss_classifier: 0.0665 (0.0664)  loss_box_reg: 0.1570 (0.1600)  loss_objectness: 0.0015 (0.0015)  loss_rpn_box_reg: 0.0043 (0.0058)  time: 1.1000  data: 0.1032  max mem: 3875
Epoch: [12]  [ 20/104]  eta: 0:01:29  lr: 0.000001  loss: 0.2113 (0.2083)  loss_classifier: 0.0605 (0.0593)  loss_box_reg: 0.1419 (0.1427)  loss_objectness: 0.0012 (0.0013)  loss_rpn_box_reg: 0.0047 (0.0051)  time: 1.0202  data: 0.0198  max mem: 3875
Epoch: [12]  [ 30/104]  eta: 0:01:17  lr: 0.000001  loss: 0.1770 (0.2075)  loss_classifier: 0.0544 (0.0600)  loss_box_reg: 0.1148 (0.1406)  loss_objectness: 0.0011 (0.0015)  loss_rpn_box_reg: 0.0047 (0.0054)  time: 1.0279  data: 0.0251  max mem: 3875
Epoch: [12]  [ 40/104]  eta: 0:01:06  lr: 0.000001  loss: 0.1848 (0.2008)  loss_classifier: 0.0544 (0.0577)  loss_box_reg: 0.1250 (0.1359)  loss_objectness: 0.0015 (0.0017)  loss_rpn_box_reg: 0.0047 (0.0053)  time: 1.0148  data: 0.0253  max mem: 3875
Epoch: [12]  [ 50/104]  eta: 0:00:55  lr: 0.000001  loss: 0.1868 (0.2007)  loss_classifier: 0.0520 (0.0573)  loss_box_reg: 0.1283 (0.1362)  loss_objectness: 0.0013 (0.0018)  loss_rpn_box_reg: 0.0041 (0.0054)  time: 0.9901  data: 0.0209  max mem: 3875
Epoch: [12]  [ 60/104]  eta: 0:00:44  lr: 0.000001  loss: 0.2124 (0.2051)  loss_classifier: 0.0576 (0.0589)  loss_box_reg: 0.1465 (0.1387)  loss_objectness: 0.0012 (0.0018)  loss_rpn_box_reg: 0.0057 (0.0057)  time: 0.9809  data: 0.0218  max mem: 3875
Epoch: [12]  [ 70/104]  eta: 0:00:34  lr: 0.000001  loss: 0.2013 (0.2020)  loss_classifier: 0.0571 (0.0577)  loss_box_reg: 0.1352 (0.1369)  loss_objectness: 0.0012 (0.0018)  loss_rpn_box_reg: 0.0057 (0.0056)  time: 0.9821  data: 0.0221  max mem: 3875
Epoch: [12]  [ 80/104]  eta: 0:00:24  lr: 0.000001  loss: 0.1894 (0.2041)  loss_classifier: 0.0489 (0.0579)  loss_box_reg: 0.1314 (0.1388)  loss_objectness: 0.0008 (0.0017)  loss_rpn_box_reg: 0.0042 (0.0057)  time: 0.9809  data: 0.0210  max mem: 3875
Epoch: [12]  [ 90/104]  eta: 0:00:14  lr: 0.000001  loss: 0.2155 (0.2062)  loss_classifier: 0.0497 (0.0582)  loss_box_reg: 0.1516 (0.1404)  loss_objectness: 0.0008 (0.0018)  loss_rpn_box_reg: 0.0044 (0.0058)  time: 0.9864  data: 0.0212  max mem: 3875
Epoch: [12]  [100/104]  eta: 0:00:04  lr: 0.000001  loss: 0.1985 (0.2077)  loss_classifier: 0.0562 (0.0584)  loss_box_reg: 0.1371 (0.1418)  loss_objectness: 0.0011 (0.0017)  loss_rpn_box_reg: 0.0056 (0.0058)  time: 0.9979  data: 0.0213  max mem: 3875
Epoch: [12]  [103/104]  eta: 0:00:01  lr: 0.000001  loss: 0.1981 (0.2069)  loss_classifier: 0.0543 (0.0583)  loss_box_reg: 0.1312 (0.1411)  loss_objectness: 0.0011 (0.0017)  loss_rpn_box_reg: 0.0056 (0.0058)  time: 0.9971  data: 0.0205  max mem: 3875
Epoch: [12] Total time: 0:01:44 (1.0093 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:36  model_time: 0.4881 (0.4881)  evaluator_time: 0.0641 (0.0641)  time: 1.4223  data: 0.8640  max mem: 3875
Test:  [25/26]  eta: 0:00:00  model_time: 0.4172 (0.4209)  evaluator_time: 0.0115 (0.0287)  time: 0.4648  data: 0.0180  max mem: 3875
Test: Total time: 0:00:13 (0.5111 s / it)
Averaged stats: model_time: 0.4172 (0.4209)  evaluator_time: 0.0115 (0.0287)
Accumulating evaluation results...
DONE (t=0.08s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.549
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.928
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.585
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.487
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.547
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.627
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.236
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.544
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.615
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.548
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.602
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.673
Epoch: [13]  [  0/104]  eta: 0:03:02  lr: 0.000001  loss: 0.2508 (0.2508)  loss_classifier: 0.0759 (0.0759)  loss_box_reg: 0.1537 (0.1537)  loss_objectness: 0.0072 (0.0072)  loss_rpn_box_reg: 0.0140 (0.0140)  time: 1.7528  data: 0.6909  max mem: 3875
Epoch: [13]  [ 10/104]  eta: 0:01:42  lr: 0.000001  loss: 0.1942 (0.2034)  loss_classifier: 0.0629 (0.0602)  loss_box_reg: 0.1263 (0.1356)  loss_objectness: 0.0011 (0.0017)  loss_rpn_box_reg: 0.0049 (0.0059)  time: 1.0934  data: 0.0841  max mem: 3875
Epoch: [13]  [ 20/104]  eta: 0:01:29  lr: 0.000001  loss: 0.2218 (0.2154)  loss_classifier: 0.0629 (0.0634)  loss_box_reg: 0.1498 (0.1432)  loss_objectness: 0.0011 (0.0019)  loss_rpn_box_reg: 0.0051 (0.0069)  time: 1.0301  data: 0.0222  max mem: 3875
Epoch: [13]  [ 30/104]  eta: 0:01:17  lr: 0.000001  loss: 0.2237 (0.2180)  loss_classifier: 0.0627 (0.0624)  loss_box_reg: 0.1510 (0.1468)  loss_objectness: 0.0017 (0.0020)  loss_rpn_box_reg: 0.0073 (0.0068)  time: 1.0236  data: 0.0205  max mem: 3875
Epoch: [13]  [ 40/104]  eta: 0:01:06  lr: 0.000001  loss: 0.2167 (0.2199)  loss_classifier: 0.0592 (0.0630)  loss_box_reg: 0.1440 (0.1485)  loss_objectness: 0.0014 (0.0018)  loss_rpn_box_reg: 0.0050 (0.0066)  time: 1.0062  data: 0.0216  max mem: 3875
Epoch: [13]  [ 50/104]  eta: 0:00:55  lr: 0.000001  loss: 0.2058 (0.2169)  loss_classifier: 0.0584 (0.0628)  loss_box_reg: 0.1418 (0.1458)  loss_objectness: 0.0010 (0.0018)  loss_rpn_box_reg: 0.0050 (0.0064)  time: 0.9935  data: 0.0228  max mem: 3875
Epoch: [13]  [ 60/104]  eta: 0:00:44  lr: 0.000001  loss: 0.1800 (0.2084)  loss_classifier: 0.0458 (0.0600)  loss_box_reg: 0.1195 (0.1408)  loss_objectness: 0.0009 (0.0017)  loss_rpn_box_reg: 0.0039 (0.0060)  time: 0.9843  data: 0.0220  max mem: 3875
Epoch: [13]  [ 70/104]  eta: 0:00:34  lr: 0.000001  loss: 0.1557 (0.2049)  loss_classifier: 0.0400 (0.0591)  loss_box_reg: 0.1138 (0.1381)  loss_objectness: 0.0008 (0.0018)  loss_rpn_box_reg: 0.0031 (0.0059)  time: 0.9763  data: 0.0208  max mem: 3875
Epoch: [13]  [ 80/104]  eta: 0:00:24  lr: 0.000001  loss: 0.1650 (0.2048)  loss_classifier: 0.0450 (0.0581)  loss_box_reg: 0.1137 (0.1389)  loss_objectness: 0.0016 (0.0019)  loss_rpn_box_reg: 0.0032 (0.0059)  time: 0.9799  data: 0.0215  max mem: 3875
Epoch: [13]  [ 90/104]  eta: 0:00:14  lr: 0.000001  loss: 0.1724 (0.2060)  loss_classifier: 0.0468 (0.0580)  loss_box_reg: 0.1282 (0.1401)  loss_objectness: 0.0014 (0.0020)  loss_rpn_box_reg: 0.0035 (0.0059)  time: 0.9912  data: 0.0218  max mem: 3875
Epoch: [13]  [100/104]  eta: 0:00:04  lr: 0.000001  loss: 0.2136 (0.2077)  loss_classifier: 0.0553 (0.0585)  loss_box_reg: 0.1544 (0.1414)  loss_objectness: 0.0011 (0.0019)  loss_rpn_box_reg: 0.0049 (0.0058)  time: 0.9947  data: 0.0200  max mem: 3875
Epoch: [13]  [103/104]  eta: 0:00:01  lr: 0.000001  loss: 0.2011 (0.2070)  loss_classifier: 0.0550 (0.0583)  loss_box_reg: 0.1402 (0.1410)  loss_objectness: 0.0011 (0.0019)  loss_rpn_box_reg: 0.0049 (0.0058)  time: 0.9969  data: 0.0197  max mem: 3875
Epoch: [13] Total time: 0:01:44 (1.0073 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:39  model_time: 0.4654 (0.4654)  evaluator_time: 0.0623 (0.0623)  time: 1.5137  data: 0.9798  max mem: 3875
Test:  [25/26]  eta: 0:00:00  model_time: 0.4157 (0.4178)  evaluator_time: 0.0113 (0.0240)  time: 0.4648  data: 0.0185  max mem: 3875
Test: Total time: 0:00:13 (0.5070 s / it)
Averaged stats: model_time: 0.4157 (0.4178)  evaluator_time: 0.0113 (0.0240)
Accumulating evaluation results...
DONE (t=0.07s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.549
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.928
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.585
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.487
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.547
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.627
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.236
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.544
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.615
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.548
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.602
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.673
Epoch: [14]  [  0/104]  eta: 0:02:57  lr: 0.000001  loss: 0.1479 (0.1479)  loss_classifier: 0.0408 (0.0408)  loss_box_reg: 0.1007 (0.1007)  loss_objectness: 0.0017 (0.0017)  loss_rpn_box_reg: 0.0047 (0.0047)  time: 1.7073  data: 0.5621  max mem: 3875
Epoch: [14]  [ 10/104]  eta: 0:01:41  lr: 0.000001  loss: 0.2075 (0.2239)  loss_classifier: 0.0596 (0.0654)  loss_box_reg: 0.1381 (0.1493)  loss_objectness: 0.0017 (0.0026)  loss_rpn_box_reg: 0.0062 (0.0066)  time: 1.0822  data: 0.0692  max mem: 3875
Epoch: [14]  [ 20/104]  eta: 0:01:28  lr: 0.000001  loss: 0.1953 (0.2120)  loss_classifier: 0.0568 (0.0602)  loss_box_reg: 0.1373 (0.1434)  loss_objectness: 0.0013 (0.0022)  loss_rpn_box_reg: 0.0049 (0.0062)  time: 1.0247  data: 0.0204  max mem: 3875
Epoch: [14]  [ 30/104]  eta: 0:01:17  lr: 0.000001  loss: 0.1721 (0.1999)  loss_classifier: 0.0488 (0.0562)  loss_box_reg: 0.1191 (0.1355)  loss_objectness: 0.0013 (0.0021)  loss_rpn_box_reg: 0.0049 (0.0061)  time: 1.0243  data: 0.0215  max mem: 3875
Epoch: [14]  [ 40/104]  eta: 0:01:06  lr: 0.000001  loss: 0.2000 (0.2106)  loss_classifier: 0.0506 (0.0585)  loss_box_reg: 0.1414 (0.1435)  loss_objectness: 0.0014 (0.0021)  loss_rpn_box_reg: 0.0059 (0.0065)  time: 1.0097  data: 0.0222  max mem: 3875
Epoch: [14]  [ 50/104]  eta: 0:00:55  lr: 0.000001  loss: 0.2065 (0.2073)  loss_classifier: 0.0590 (0.0581)  loss_box_reg: 0.1524 (0.1412)  loss_objectness: 0.0014 (0.0019)  loss_rpn_box_reg: 0.0051 (0.0061)  time: 0.9892  data: 0.0216  max mem: 3875
Epoch: [14]  [ 60/104]  eta: 0:00:44  lr: 0.000001  loss: 0.1814 (0.2034)  loss_classifier: 0.0509 (0.0570)  loss_box_reg: 0.1188 (0.1389)  loss_objectness: 0.0006 (0.0019)  loss_rpn_box_reg: 0.0037 (0.0057)  time: 0.9771  data: 0.0212  max mem: 3875
Epoch: [14]  [ 70/104]  eta: 0:00:34  lr: 0.000001  loss: 0.1878 (0.2071)  loss_classifier: 0.0516 (0.0582)  loss_box_reg: 0.1317 (0.1411)  loss_objectness: 0.0015 (0.0019)  loss_rpn_box_reg: 0.0042 (0.0059)  time: 0.9815  data: 0.0226  max mem: 3875
Epoch: [14]  [ 80/104]  eta: 0:00:24  lr: 0.000001  loss: 0.1900 (0.2046)  loss_classifier: 0.0562 (0.0580)  loss_box_reg: 0.1317 (0.1390)  loss_objectness: 0.0015 (0.0019)  loss_rpn_box_reg: 0.0052 (0.0057)  time: 0.9864  data: 0.0240  max mem: 3875
Epoch: [14]  [ 90/104]  eta: 0:00:14  lr: 0.000001  loss: 0.1885 (0.2033)  loss_classifier: 0.0524 (0.0573)  loss_box_reg: 0.1268 (0.1385)  loss_objectness: 0.0014 (0.0019)  loss_rpn_box_reg: 0.0040 (0.0056)  time: 0.9856  data: 0.0220  max mem: 3875
Epoch: [14]  [100/104]  eta: 0:00:04  lr: 0.000001  loss: 0.1979 (0.2062)  loss_classifier: 0.0538 (0.0581)  loss_box_reg: 0.1322 (0.1403)  loss_objectness: 0.0014 (0.0019)  loss_rpn_box_reg: 0.0053 (0.0059)  time: 0.9956  data: 0.0202  max mem: 3875
Epoch: [14]  [103/104]  eta: 0:00:01  lr: 0.000001  loss: 0.2031 (0.2073)  loss_classifier: 0.0576 (0.0582)  loss_box_reg: 0.1392 (0.1414)  loss_objectness: 0.0015 (0.0019)  loss_rpn_box_reg: 0.0056 (0.0058)  time: 0.9960  data: 0.0196  max mem: 3875
Epoch: [14] Total time: 0:01:44 (1.0063 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:46  model_time: 0.5431 (0.5431)  evaluator_time: 0.2347 (0.2347)  time: 1.8017  data: 0.9980  max mem: 3875
Test:  [25/26]  eta: 0:00:00  model_time: 0.4153 (0.4221)  evaluator_time: 0.0111 (0.0321)  time: 0.4574  data: 0.0175  max mem: 3875
Test: Total time: 0:00:13 (0.5213 s / it)
Averaged stats: model_time: 0.4153 (0.4221)  evaluator_time: 0.0111 (0.0321)
Accumulating evaluation results...
DONE (t=0.08s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.549
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.928
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.585
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.487
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.547
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.627
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.236
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.544
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.615
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.548
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.602
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.673

# to train on gpu if selected.
device = torch.device('cuda') if torch.cuda.is_available() else torch.device('cpu')


num_classes = 11

# get the model using our helper function
model = get_object_detection_model(num_classes)

# move model to the right device
model.to(device)

# construct an optimizer
params = [p for p in model.parameters() if p.requires_grad]
optimizer = torch.optim.SGD(params, lr=0.005,
                            momentum=0.9, weight_decay=0.0005)

# and a learning rate scheduler which decreases the learning rate by
# 10x every 3 epochs
lr_scheduler = torch.optim.lr_scheduler.StepLR(optimizer,
                                               step_size=3,
                                               gamma=0.1)

# training for 10 epochs
num_epochs = 20

for epoch in range(num_epochs):
    # training for one epoch
    train_one_epoch(model, optimizer, data_loader, device, epoch, print_freq=10)
    # update the learning rate
    lr_scheduler.step()
    # evaluate on the test dataset
    evaluate(model, data_loader_test, device=device)

Epoch: [0]  [  0/104]  eta: 0:11:35  lr: 0.000053  loss: 2.7235 (2.7235)  loss_classifier: 2.3228 (2.3228)  loss_box_reg: 0.2357 (0.2357)  loss_objectness: 0.1499 (0.1499)  loss_rpn_box_reg: 0.0151 (0.0151)  time: 6.6916  data: 4.5670  max mem: 4198
Epoch: [0]  [ 10/104]  eta: 0:02:26  lr: 0.000538  loss: 2.5347 (2.2935)  loss_classifier: 1.8309 (1.6350)  loss_box_reg: 0.3244 (0.3192)  loss_objectness: 0.2873 (0.3125)  loss_rpn_box_reg: 0.0241 (0.0267)  time: 1.5552  data: 0.4836  max mem: 4355
Epoch: [0]  [ 20/104]  eta: 0:01:51  lr: 0.001023  loss: 1.2679 (1.6503)  loss_classifier: 0.7718 (1.1099)  loss_box_reg: 0.2971 (0.3030)  loss_objectness: 0.1352 (0.2163)  loss_rpn_box_reg: 0.0162 (0.0212)  time: 1.0557  data: 0.0636  max mem: 4355
Epoch: [0]  [ 30/104]  eta: 0:01:30  lr: 0.001508  loss: 0.9645 (1.4740)  loss_classifier: 0.4899 (0.9395)  loss_box_reg: 0.3449 (0.3378)  loss_objectness: 0.0906 (0.1763)  loss_rpn_box_reg: 0.0142 (0.0204)  time: 1.0487  data: 0.0617  max mem: 4355
Epoch: [0]  [ 40/104]  eta: 0:01:17  lr: 0.001993  loss: 0.9903 (1.3530)  loss_classifier: 0.4853 (0.8226)  loss_box_reg: 0.4206 (0.3615)  loss_objectness: 0.0740 (0.1492)  loss_rpn_box_reg: 0.0139 (0.0198)  time: 1.0871  data: 0.1443  max mem: 4355
Epoch: [0]  [ 50/104]  eta: 0:01:02  lr: 0.002478  loss: 0.9366 (1.2888)  loss_classifier: 0.4267 (0.7586)  loss_box_reg: 0.4388 (0.3795)  loss_objectness: 0.0425 (0.1312)  loss_rpn_box_reg: 0.0131 (0.0196)  time: 1.0690  data: 0.1398  max mem: 4355
Epoch: [0]  [ 60/104]  eta: 0:00:50  lr: 0.002963  loss: 0.8294 (1.2162)  loss_classifier: 0.3904 (0.6984)  loss_box_reg: 0.3764 (0.3836)  loss_objectness: 0.0404 (0.1156)  loss_rpn_box_reg: 0.0120 (0.0186)  time: 1.0098  data: 0.0847  max mem: 4355
Epoch: [0]  [ 70/104]  eta: 0:00:39  lr: 0.003448  loss: 0.7882 (1.1589)  loss_classifier: 0.3670 (0.6539)  loss_box_reg: 0.3694 (0.3828)  loss_objectness: 0.0283 (0.1033)  loss_rpn_box_reg: 0.0120 (0.0189)  time: 1.1152  data: 0.1922  max mem: 4355
Epoch: [0]  [ 80/104]  eta: 0:00:27  lr: 0.003933  loss: 0.8186 (1.1105)  loss_classifier: 0.3795 (0.6147)  loss_box_reg: 0.3904 (0.3820)  loss_objectness: 0.0218 (0.0949)  loss_rpn_box_reg: 0.0144 (0.0189)  time: 1.1340  data: 0.2052  max mem: 4355
Epoch: [0]  [ 90/104]  eta: 0:00:15  lr: 0.004418  loss: 0.7866 (1.0753)  loss_classifier: 0.3404 (0.5858)  loss_box_reg: 0.3858 (0.3820)  loss_objectness: 0.0246 (0.0889)  loss_rpn_box_reg: 0.0124 (0.0186)  time: 1.1019  data: 0.1557  max mem: 4355
Epoch: [0]  [100/104]  eta: 0:00:04  lr: 0.004903  loss: 0.7400 (1.0362)  loss_classifier: 0.2854 (0.5545)  loss_box_reg: 0.3585 (0.3797)  loss_objectness: 0.0327 (0.0834)  loss_rpn_box_reg: 0.0130 (0.0185)  time: 1.2998  data: 0.3455  max mem: 4355
Epoch: [0]  [103/104]  eta: 0:00:01  lr: 0.005000  loss: 0.7400 (1.0289)  loss_classifier: 0.2854 (0.5487)  loss_box_reg: 0.3585 (0.3797)  loss_objectness: 0.0327 (0.0820)  loss_rpn_box_reg: 0.0131 (0.0185)  time: 1.2825  data: 0.3306  max mem: 4355
Epoch: [0] Total time: 0:02:01 (1.1669 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:32  model_time: 0.4620 (0.4620)  evaluator_time: 0.0388 (0.0388)  time: 1.2388  data: 0.7165  max mem: 4355
Test:  [25/26]  eta: 0:00:00  model_time: 0.4043 (0.4065)  evaluator_time: 0.0297 (0.0362)  time: 0.4647  data: 0.0182  max mem: 4355
Test: Total time: 0:00:12 (0.4976 s / it)
Averaged stats: model_time: 0.4043 (0.4065)  evaluator_time: 0.0297 (0.0362)
Accumulating evaluation results...
DONE (t=0.20s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.247
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.594
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.160
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.191
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.278
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.268
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.118
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.319
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.391
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.330
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.386
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.330
Epoch: [1]  [  0/104]  eta: 0:03:02  lr: 0.005000  loss: 0.7628 (0.7628)  loss_classifier: 0.2548 (0.2548)  loss_box_reg: 0.4684 (0.4684)  loss_objectness: 0.0259 (0.0259)  loss_rpn_box_reg: 0.0136 (0.0136)  time: 1.7539  data: 0.7692  max mem: 4355
Epoch: [1]  [ 10/104]  eta: 0:01:44  lr: 0.005000  loss: 0.5767 (0.5778)  loss_classifier: 0.2348 (0.2165)  loss_box_reg: 0.3276 (0.3243)  loss_objectness: 0.0185 (0.0217)  loss_rpn_box_reg: 0.0118 (0.0152)  time: 1.1133  data: 0.0867  max mem: 4355
Epoch: [1]  [ 20/104]  eta: 0:01:29  lr: 0.005000  loss: 0.4675 (0.5555)  loss_classifier: 0.1614 (0.2031)  loss_box_reg: 0.3077 (0.3187)  loss_objectness: 0.0104 (0.0195)  loss_rpn_box_reg: 0.0100 (0.0142)  time: 1.0367  data: 0.0259  max mem: 4355
Epoch: [1]  [ 30/104]  eta: 0:01:18  lr: 0.005000  loss: 0.4675 (0.5371)  loss_classifier: 0.1614 (0.2008)  loss_box_reg: 0.2925 (0.3037)  loss_objectness: 0.0117 (0.0190)  loss_rpn_box_reg: 0.0091 (0.0137)  time: 1.0227  data: 0.0331  max mem: 4355
Epoch: [1]  [ 40/104]  eta: 0:01:06  lr: 0.005000  loss: 0.5261 (0.5451)  loss_classifier: 0.1940 (0.2019)  loss_box_reg: 0.2942 (0.3109)  loss_objectness: 0.0136 (0.0184)  loss_rpn_box_reg: 0.0108 (0.0139)  time: 0.9918  data: 0.0268  max mem: 4355
Epoch: [1]  [ 50/104]  eta: 0:00:54  lr: 0.005000  loss: 0.5389 (0.5465)  loss_classifier: 0.1940 (0.1996)  loss_box_reg: 0.3290 (0.3142)  loss_objectness: 0.0127 (0.0181)  loss_rpn_box_reg: 0.0151 (0.0147)  time: 0.9576  data: 0.0213  max mem: 4355
Epoch: [1]  [ 60/104]  eta: 0:00:44  lr: 0.005000  loss: 0.4767 (0.5370)  loss_classifier: 0.1476 (0.1920)  loss_box_reg: 0.2888 (0.3127)  loss_objectness: 0.0125 (0.0177)  loss_rpn_box_reg: 0.0153 (0.0146)  time: 0.9472  data: 0.0210  max mem: 4355
Epoch: [1]  [ 70/104]  eta: 0:00:33  lr: 0.005000  loss: 0.4507 (0.5306)  loss_classifier: 0.1376 (0.1887)  loss_box_reg: 0.2775 (0.3097)  loss_objectness: 0.0130 (0.0179)  loss_rpn_box_reg: 0.0121 (0.0143)  time: 0.9431  data: 0.0207  max mem: 4355
Epoch: [1]  [ 80/104]  eta: 0:00:23  lr: 0.005000  loss: 0.4959 (0.5238)  loss_classifier: 0.1621 (0.1862)  loss_box_reg: 0.3003 (0.3065)  loss_objectness: 0.0094 (0.0173)  loss_rpn_box_reg: 0.0118 (0.0138)  time: 0.9468  data: 0.0213  max mem: 4355
Epoch: [1]  [ 90/104]  eta: 0:00:13  lr: 0.005000  loss: 0.4317 (0.5092)  loss_classifier: 0.1319 (0.1791)  loss_box_reg: 0.2675 (0.2997)  loss_objectness: 0.0085 (0.0166)  loss_rpn_box_reg: 0.0118 (0.0137)  time: 0.9522  data: 0.0212  max mem: 4355
Epoch: [1]  [100/104]  eta: 0:00:03  lr: 0.005000  loss: 0.3731 (0.4952)  loss_classifier: 0.1139 (0.1720)  loss_box_reg: 0.2457 (0.2939)  loss_objectness: 0.0069 (0.0158)  loss_rpn_box_reg: 0.0106 (0.0135)  time: 0.9506  data: 0.0192  max mem: 4355
Epoch: [1]  [103/104]  eta: 0:00:00  lr: 0.005000  loss: 0.3872 (0.4931)  loss_classifier: 0.1150 (0.1705)  loss_box_reg: 0.2577 (0.2935)  loss_objectness: 0.0065 (0.0158)  loss_rpn_box_reg: 0.0086 (0.0133)  time: 0.9518  data: 0.0192  max mem: 4355
Epoch: [1] Total time: 0:01:42 (0.9831 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:41  model_time: 0.5403 (0.5403)  evaluator_time: 0.0770 (0.0770)  time: 1.6116  data: 0.9682  max mem: 4355
Test:  [25/26]  eta: 0:00:00  model_time: 0.4011 (0.4075)  evaluator_time: 0.0145 (0.0261)  time: 0.4491  data: 0.0177  max mem: 4355
Test: Total time: 0:00:13 (0.5019 s / it)
Averaged stats: model_time: 0.4011 (0.4075)  evaluator_time: 0.0145 (0.0261)
Accumulating evaluation results...
DONE (t=0.23s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.435
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.840
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.384
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.347
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.425
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.361
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.183
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.453
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.522
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.465
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.509
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.494
Epoch: [2]  [  0/104]  eta: 0:03:53  lr: 0.005000  loss: 0.4320 (0.4320)  loss_classifier: 0.1283 (0.1283)  loss_box_reg: 0.2721 (0.2721)  loss_objectness: 0.0149 (0.0149)  loss_rpn_box_reg: 0.0168 (0.0168)  time: 2.2433  data: 1.1228  max mem: 4355
Epoch: [2]  [ 10/104]  eta: 0:01:41  lr: 0.005000  loss: 0.4269 (0.3874)  loss_classifier: 0.1283 (0.1159)  loss_box_reg: 0.2525 (0.2509)  loss_objectness: 0.0081 (0.0098)  loss_rpn_box_reg: 0.0102 (0.0107)  time: 1.0842  data: 0.1168  max mem: 4355
Epoch: [2]  [ 20/104]  eta: 0:01:27  lr: 0.005000  loss: 0.4396 (0.4110)  loss_classifier: 0.1152 (0.1137)  loss_box_reg: 0.2905 (0.2789)  loss_objectness: 0.0060 (0.0083)  loss_rpn_box_reg: 0.0094 (0.0100)  time: 0.9769  data: 0.0177  max mem: 4355
Epoch: [2]  [ 30/104]  eta: 0:01:15  lr: 0.005000  loss: 0.4396 (0.3977)  loss_classifier: 0.1056 (0.1137)  loss_box_reg: 0.2872 (0.2660)  loss_objectness: 0.0051 (0.0076)  loss_rpn_box_reg: 0.0094 (0.0102)  time: 0.9839  data: 0.0207  max mem: 4355
Epoch: [2]  [ 40/104]  eta: 0:01:04  lr: 0.005000  loss: 0.3588 (0.3833)  loss_classifier: 0.1041 (0.1127)  loss_box_reg: 0.2210 (0.2532)  loss_objectness: 0.0044 (0.0069)  loss_rpn_box_reg: 0.0103 (0.0104)  time: 0.9737  data: 0.0212  max mem: 4355
Epoch: [2]  [ 50/104]  eta: 0:00:53  lr: 0.005000  loss: 0.3516 (0.3717)  loss_classifier: 0.0992 (0.1090)  loss_box_reg: 0.2057 (0.2460)  loss_objectness: 0.0034 (0.0064)  loss_rpn_box_reg: 0.0103 (0.0103)  time: 0.9580  data: 0.0201  max mem: 4355
Epoch: [2]  [ 60/104]  eta: 0:00:43  lr: 0.005000  loss: 0.3467 (0.3702)  loss_classifier: 0.0824 (0.1086)  loss_box_reg: 0.2244 (0.2451)  loss_objectness: 0.0035 (0.0063)  loss_rpn_box_reg: 0.0079 (0.0101)  time: 0.9483  data: 0.0194  max mem: 4355
Epoch: [2]  [ 70/104]  eta: 0:00:33  lr: 0.005000  loss: 0.3447 (0.3655)  loss_classifier: 0.0935 (0.1070)  loss_box_reg: 0.2187 (0.2422)  loss_objectness: 0.0051 (0.0064)  loss_rpn_box_reg: 0.0079 (0.0099)  time: 0.9445  data: 0.0193  max mem: 4355
Epoch: [2]  [ 80/104]  eta: 0:00:23  lr: 0.005000  loss: 0.3687 (0.3694)  loss_classifier: 0.1029 (0.1079)  loss_box_reg: 0.2330 (0.2452)  loss_objectness: 0.0051 (0.0061)  loss_rpn_box_reg: 0.0077 (0.0102)  time: 0.9436  data: 0.0198  max mem: 4355
Epoch: [2]  [ 90/104]  eta: 0:00:13  lr: 0.005000  loss: 0.3849 (0.3707)  loss_classifier: 0.1032 (0.1066)  loss_box_reg: 0.2601 (0.2481)  loss_objectness: 0.0039 (0.0060)  loss_rpn_box_reg: 0.0075 (0.0101)  time: 0.9504  data: 0.0211  max mem: 4355
Epoch: [2]  [100/104]  eta: 0:00:03  lr: 0.005000  loss: 0.3699 (0.3734)  loss_classifier: 0.0956 (0.1076)  loss_box_reg: 0.2644 (0.2494)  loss_objectness: 0.0047 (0.0062)  loss_rpn_box_reg: 0.0064 (0.0101)  time: 0.9547  data: 0.0202  max mem: 4355
Epoch: [2]  [103/104]  eta: 0:00:00  lr: 0.005000  loss: 0.3719 (0.3723)  loss_classifier: 0.0956 (0.1077)  loss_box_reg: 0.2644 (0.2483)  loss_objectness: 0.0061 (0.0062)  loss_rpn_box_reg: 0.0085 (0.0101)  time: 0.9520  data: 0.0186  max mem: 4355
Epoch: [2] Total time: 0:01:41 (0.9723 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:27  model_time: 0.4974 (0.4974)  evaluator_time: 0.0442 (0.0442)  time: 1.0693  data: 0.5080  max mem: 4355
Test:  [25/26]  eta: 0:00:00  model_time: 0.4038 (0.4073)  evaluator_time: 0.0153 (0.0251)  time: 0.4540  data: 0.0192  max mem: 4355
Test: Total time: 0:00:12 (0.4815 s / it)
Averaged stats: model_time: 0.4038 (0.4073)  evaluator_time: 0.0153 (0.0251)
Accumulating evaluation results...
DONE (t=0.11s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.465
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.878
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.438
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.336
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.465
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.374
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.196
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.481
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.549
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.491
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.554
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.591
Epoch: [3]  [  0/104]  eta: 0:02:45  lr: 0.000500  loss: 0.3921 (0.3921)  loss_classifier: 0.1305 (0.1305)  loss_box_reg: 0.2452 (0.2452)  loss_objectness: 0.0076 (0.0076)  loss_rpn_box_reg: 0.0089 (0.0089)  time: 1.5954  data: 0.5706  max mem: 4355
Epoch: [3]  [ 10/104]  eta: 0:01:36  lr: 0.000500  loss: 0.3498 (0.3337)  loss_classifier: 0.0860 (0.0953)  loss_box_reg: 0.2140 (0.2222)  loss_objectness: 0.0032 (0.0046)  loss_rpn_box_reg: 0.0089 (0.0117)  time: 1.0275  data: 0.0692  max mem: 4355
Epoch: [3]  [ 20/104]  eta: 0:01:24  lr: 0.000500  loss: 0.2819 (0.2992)  loss_classifier: 0.0810 (0.0836)  loss_box_reg: 0.1910 (0.2023)  loss_objectness: 0.0030 (0.0040)  loss_rpn_box_reg: 0.0067 (0.0092)  time: 0.9759  data: 0.0190  max mem: 4355
Epoch: [3]  [ 30/104]  eta: 0:01:13  lr: 0.000500  loss: 0.2535 (0.3029)  loss_classifier: 0.0745 (0.0833)  loss_box_reg: 0.1924 (0.2066)  loss_objectness: 0.0030 (0.0042)  loss_rpn_box_reg: 0.0057 (0.0088)  time: 0.9776  data: 0.0192  max mem: 4355
Epoch: [3]  [ 40/104]  eta: 0:01:03  lr: 0.000500  loss: 0.2572 (0.2946)  loss_classifier: 0.0745 (0.0822)  loss_box_reg: 0.1924 (0.1994)  loss_objectness: 0.0036 (0.0043)  loss_rpn_box_reg: 0.0075 (0.0086)  time: 0.9670  data: 0.0197  max mem: 4355
Epoch: [3]  [ 50/104]  eta: 0:00:52  lr: 0.000500  loss: 0.2402 (0.2868)  loss_classifier: 0.0735 (0.0808)  loss_box_reg: 0.1591 (0.1937)  loss_objectness: 0.0027 (0.0041)  loss_rpn_box_reg: 0.0062 (0.0083)  time: 0.9513  data: 0.0189  max mem: 4355
Epoch: [3]  [ 60/104]  eta: 0:00:42  lr: 0.000500  loss: 0.2573 (0.2853)  loss_classifier: 0.0810 (0.0809)  loss_box_reg: 0.1642 (0.1923)  loss_objectness: 0.0024 (0.0039)  loss_rpn_box_reg: 0.0059 (0.0082)  time: 0.9440  data: 0.0187  max mem: 4355
Epoch: [3]  [ 70/104]  eta: 0:00:32  lr: 0.000500  loss: 0.2573 (0.2814)  loss_classifier: 0.0799 (0.0798)  loss_box_reg: 0.1760 (0.1898)  loss_objectness: 0.0023 (0.0037)  loss_rpn_box_reg: 0.0053 (0.0080)  time: 0.9472  data: 0.0211  max mem: 4355
Epoch: [3]  [ 80/104]  eta: 0:00:23  lr: 0.000500  loss: 0.2383 (0.2781)  loss_classifier: 0.0648 (0.0788)  loss_box_reg: 0.1659 (0.1874)  loss_objectness: 0.0023 (0.0036)  loss_rpn_box_reg: 0.0047 (0.0083)  time: 0.9517  data: 0.0217  max mem: 4355
Epoch: [3]  [ 90/104]  eta: 0:00:13  lr: 0.000500  loss: 0.2320 (0.2716)  loss_classifier: 0.0633 (0.0769)  loss_box_reg: 0.1460 (0.1835)  loss_objectness: 0.0011 (0.0033)  loss_rpn_box_reg: 0.0047 (0.0079)  time: 0.9578  data: 0.0221  max mem: 4355
Epoch: [3]  [100/104]  eta: 0:00:03  lr: 0.000500  loss: 0.2102 (0.2685)  loss_classifier: 0.0623 (0.0763)  loss_box_reg: 0.1523 (0.1811)  loss_objectness: 0.0012 (0.0032)  loss_rpn_box_reg: 0.0048 (0.0079)  time: 0.9575  data: 0.0216  max mem: 4355
Epoch: [3]  [103/104]  eta: 0:00:00  lr: 0.000500  loss: 0.2446 (0.2679)  loss_classifier: 0.0623 (0.0763)  loss_box_reg: 0.1544 (0.1807)  loss_objectness: 0.0012 (0.0032)  loss_rpn_box_reg: 0.0048 (0.0078)  time: 0.9498  data: 0.0192  max mem: 4355
Epoch: [3] Total time: 0:01:40 (0.9662 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:30  model_time: 0.5140 (0.5140)  evaluator_time: 0.0180 (0.0180)  time: 1.1602  data: 0.6156  max mem: 4355
Test:  [25/26]  eta: 0:00:00  model_time: 0.4073 (0.4099)  evaluator_time: 0.0152 (0.0249)  time: 0.4665  data: 0.0246  max mem: 4355
Test: Total time: 0:00:12 (0.4921 s / it)
Averaged stats: model_time: 0.4073 (0.4099)  evaluator_time: 0.0152 (0.0249)
Accumulating evaluation results...
DONE (t=0.08s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.530
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.887
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.576
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.359
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.532
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.468
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.217
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.530
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.602
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.422
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.599
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.515
Epoch: [4]  [  0/104]  eta: 0:02:48  lr: 0.000500  loss: 0.1289 (0.1289)  loss_classifier: 0.0386 (0.0386)  loss_box_reg: 0.0862 (0.0862)  loss_objectness: 0.0019 (0.0019)  loss_rpn_box_reg: 0.0022 (0.0022)  time: 1.6187  data: 0.5553  max mem: 4355
Epoch: [4]  [ 10/104]  eta: 0:01:37  lr: 0.000500  loss: 0.2817 (0.2439)  loss_classifier: 0.0818 (0.0724)  loss_box_reg: 0.1687 (0.1613)  loss_objectness: 0.0019 (0.0030)  loss_rpn_box_reg: 0.0061 (0.0072)  time: 1.0355  data: 0.0702  max mem: 4355
Epoch: [4]  [ 20/104]  eta: 0:01:24  lr: 0.000500  loss: 0.2388 (0.2390)  loss_classifier: 0.0657 (0.0695)  loss_box_reg: 0.1585 (0.1599)  loss_objectness: 0.0019 (0.0031)  loss_rpn_box_reg: 0.0061 (0.0066)  time: 0.9768  data: 0.0204  max mem: 4355
Epoch: [4]  [ 30/104]  eta: 0:01:13  lr: 0.000500  loss: 0.2285 (0.2361)  loss_classifier: 0.0645 (0.0672)  loss_box_reg: 0.1585 (0.1598)  loss_objectness: 0.0020 (0.0028)  loss_rpn_box_reg: 0.0052 (0.0063)  time: 0.9769  data: 0.0205  max mem: 4355
Epoch: [4]  [ 40/104]  eta: 0:01:03  lr: 0.000500  loss: 0.2079 (0.2274)  loss_classifier: 0.0566 (0.0645)  loss_box_reg: 0.1348 (0.1541)  loss_objectness: 0.0016 (0.0026)  loss_rpn_box_reg: 0.0038 (0.0062)  time: 0.9729  data: 0.0222  max mem: 4355
Epoch: [4]  [ 50/104]  eta: 0:00:53  lr: 0.000500  loss: 0.2170 (0.2308)  loss_classifier: 0.0616 (0.0657)  loss_box_reg: 0.1520 (0.1558)  loss_objectness: 0.0017 (0.0025)  loss_rpn_box_reg: 0.0043 (0.0068)  time: 0.9586  data: 0.0205  max mem: 4355
Epoch: [4]  [ 60/104]  eta: 0:00:42  lr: 0.000500  loss: 0.2221 (0.2337)  loss_classifier: 0.0697 (0.0672)  loss_box_reg: 0.1547 (0.1569)  loss_objectness: 0.0021 (0.0026)  loss_rpn_box_reg: 0.0051 (0.0071)  time: 0.9470  data: 0.0189  max mem: 4355
Epoch: [4]  [ 70/104]  eta: 0:00:33  lr: 0.000500  loss: 0.2253 (0.2370)  loss_classifier: 0.0692 (0.0679)  loss_box_reg: 0.1447 (0.1593)  loss_objectness: 0.0023 (0.0026)  loss_rpn_box_reg: 0.0056 (0.0072)  time: 0.9469  data: 0.0213  max mem: 4355
Epoch: [4]  [ 80/104]  eta: 0:00:23  lr: 0.000500  loss: 0.2193 (0.2386)  loss_classifier: 0.0669 (0.0676)  loss_box_reg: 0.1486 (0.1612)  loss_objectness: 0.0019 (0.0026)  loss_rpn_box_reg: 0.0056 (0.0072)  time: 0.9489  data: 0.0220  max mem: 4355
Epoch: [4]  [ 90/104]  eta: 0:00:13  lr: 0.000500  loss: 0.2061 (0.2377)  loss_classifier: 0.0658 (0.0671)  loss_box_reg: 0.1486 (0.1609)  loss_objectness: 0.0014 (0.0026)  loss_rpn_box_reg: 0.0054 (0.0071)  time: 0.9513  data: 0.0213  max mem: 4355
Epoch: [4]  [100/104]  eta: 0:00:03  lr: 0.000500  loss: 0.2612 (0.2428)  loss_classifier: 0.0776 (0.0687)  loss_box_reg: 0.1740 (0.1644)  loss_objectness: 0.0019 (0.0026)  loss_rpn_box_reg: 0.0062 (0.0072)  time: 0.9562  data: 0.0201  max mem: 4355
Epoch: [4]  [103/104]  eta: 0:00:00  lr: 0.000500  loss: 0.2450 (0.2410)  loss_classifier: 0.0685 (0.0682)  loss_box_reg: 0.1671 (0.1631)  loss_objectness: 0.0019 (0.0026)  loss_rpn_box_reg: 0.0052 (0.0071)  time: 0.9538  data: 0.0189  max mem: 4355
Epoch: [4] Total time: 0:01:40 (0.9673 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:34  model_time: 0.4584 (0.4584)  evaluator_time: 0.0162 (0.0162)  time: 1.3111  data: 0.8223  max mem: 4355
Test:  [25/26]  eta: 0:00:00  model_time: 0.4042 (0.4074)  evaluator_time: 0.0125 (0.0184)  time: 0.4521  data: 0.0208  max mem: 4355
Test: Total time: 0:00:12 (0.4885 s / it)
Averaged stats: model_time: 0.4042 (0.4074)  evaluator_time: 0.0125 (0.0184)
Accumulating evaluation results...
DONE (t=0.08s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.534
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.887
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.600
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.371
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.531
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.464
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.215
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.526
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.602
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.427
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.598
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.497
Epoch: [5]  [  0/104]  eta: 0:02:52  lr: 0.000500  loss: 0.2231 (0.2231)  loss_classifier: 0.0715 (0.0715)  loss_box_reg: 0.1476 (0.1476)  loss_objectness: 0.0005 (0.0005)  loss_rpn_box_reg: 0.0035 (0.0035)  time: 1.6560  data: 0.6650  max mem: 4355
Epoch: [5]  [ 10/104]  eta: 0:01:37  lr: 0.000500  loss: 0.2231 (0.2124)  loss_classifier: 0.0647 (0.0623)  loss_box_reg: 0.1483 (0.1440)  loss_objectness: 0.0017 (0.0016)  loss_rpn_box_reg: 0.0043 (0.0046)  time: 1.0365  data: 0.0794  max mem: 4355
Epoch: [5]  [ 20/104]  eta: 0:01:25  lr: 0.000500  loss: 0.2201 (0.2191)  loss_classifier: 0.0637 (0.0628)  loss_box_reg: 0.1483 (0.1480)  loss_objectness: 0.0018 (0.0020)  loss_rpn_box_reg: 0.0050 (0.0063)  time: 0.9815  data: 0.0213  max mem: 4355
Epoch: [5]  [ 30/104]  eta: 0:01:13  lr: 0.000500  loss: 0.2193 (0.2201)  loss_classifier: 0.0611 (0.0623)  loss_box_reg: 0.1514 (0.1493)  loss_objectness: 0.0018 (0.0021)  loss_rpn_box_reg: 0.0065 (0.0064)  time: 0.9794  data: 0.0212  max mem: 4355
Epoch: [5]  [ 40/104]  eta: 0:01:03  lr: 0.000500  loss: 0.2193 (0.2213)  loss_classifier: 0.0586 (0.0630)  loss_box_reg: 0.1522 (0.1496)  loss_objectness: 0.0013 (0.0020)  loss_rpn_box_reg: 0.0053 (0.0067)  time: 0.9685  data: 0.0215  max mem: 4355
Epoch: [5]  [ 50/104]  eta: 0:00:53  lr: 0.000500  loss: 0.2117 (0.2212)  loss_classifier: 0.0611 (0.0635)  loss_box_reg: 0.1343 (0.1490)  loss_objectness: 0.0014 (0.0019)  loss_rpn_box_reg: 0.0045 (0.0068)  time: 0.9616  data: 0.0218  max mem: 4355
Epoch: [5]  [ 60/104]  eta: 0:00:43  lr: 0.000500  loss: 0.2117 (0.2223)  loss_classifier: 0.0637 (0.0641)  loss_box_reg: 0.1343 (0.1496)  loss_objectness: 0.0016 (0.0020)  loss_rpn_box_reg: 0.0063 (0.0066)  time: 0.9495  data: 0.0211  max mem: 4355
Epoch: [5]  [ 70/104]  eta: 0:00:33  lr: 0.000500  loss: 0.2423 (0.2266)  loss_classifier: 0.0665 (0.0650)  loss_box_reg: 0.1721 (0.1528)  loss_objectness: 0.0026 (0.0022)  loss_rpn_box_reg: 0.0067 (0.0067)  time: 0.9458  data: 0.0211  max mem: 4355
Epoch: [5]  [ 80/104]  eta: 0:00:23  lr: 0.000500  loss: 0.2581 (0.2314)  loss_classifier: 0.0672 (0.0657)  loss_box_reg: 0.1741 (0.1566)  loss_objectness: 0.0023 (0.0021)  loss_rpn_box_reg: 0.0065 (0.0070)  time: 0.9506  data: 0.0222  max mem: 4355
Epoch: [5]  [ 90/104]  eta: 0:00:13  lr: 0.000500  loss: 0.2467 (0.2312)  loss_classifier: 0.0638 (0.0654)  loss_box_reg: 0.1741 (0.1568)  loss_objectness: 0.0016 (0.0021)  loss_rpn_box_reg: 0.0052 (0.0068)  time: 0.9534  data: 0.0226  max mem: 4355
Epoch: [5]  [100/104]  eta: 0:00:03  lr: 0.000500  loss: 0.2264 (0.2324)  loss_classifier: 0.0638 (0.0659)  loss_box_reg: 0.1560 (0.1573)  loss_objectness: 0.0016 (0.0022)  loss_rpn_box_reg: 0.0061 (0.0069)  time: 0.9632  data: 0.0234  max mem: 4355
Epoch: [5]  [103/104]  eta: 0:00:00  lr: 0.000500  loss: 0.2264 (0.2302)  loss_classifier: 0.0638 (0.0653)  loss_box_reg: 0.1560 (0.1559)  loss_objectness: 0.0017 (0.0022)  loss_rpn_box_reg: 0.0062 (0.0068)  time: 0.9634  data: 0.0232  max mem: 4355
Epoch: [5] Total time: 0:01:40 (0.9698 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:31  model_time: 0.4610 (0.4610)  evaluator_time: 0.0177 (0.0177)  time: 1.2052  data: 0.7138  max mem: 4355
Test:  [25/26]  eta: 0:00:00  model_time: 0.4044 (0.4084)  evaluator_time: 0.0100 (0.0197)  time: 0.4468  data: 0.0189  max mem: 4355
Test: Total time: 0:00:12 (0.4852 s / it)
Averaged stats: model_time: 0.4044 (0.4084)  evaluator_time: 0.0100 (0.0197)
Accumulating evaluation results...
DONE (t=0.07s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.530
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.893
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.580
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.366
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.532
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.460
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.223
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.529
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.602
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.429
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.604
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.497
Epoch: [6]  [  0/104]  eta: 0:03:02  lr: 0.000050  loss: 0.2072 (0.2072)  loss_classifier: 0.0535 (0.0535)  loss_box_reg: 0.1461 (0.1461)  loss_objectness: 0.0027 (0.0027)  loss_rpn_box_reg: 0.0049 (0.0049)  time: 1.7548  data: 0.7822  max mem: 4355
Epoch: [6]  [ 10/104]  eta: 0:01:37  lr: 0.000050  loss: 0.2284 (0.2237)  loss_classifier: 0.0589 (0.0612)  loss_box_reg: 0.1529 (0.1548)  loss_objectness: 0.0017 (0.0018)  loss_rpn_box_reg: 0.0059 (0.0059)  time: 1.0415  data: 0.0879  max mem: 4355
Epoch: [6]  [ 20/104]  eta: 0:01:25  lr: 0.000050  loss: 0.2352 (0.2425)  loss_classifier: 0.0629 (0.0678)  loss_box_reg: 0.1653 (0.1652)  loss_objectness: 0.0017 (0.0022)  loss_rpn_box_reg: 0.0068 (0.0073)  time: 0.9772  data: 0.0194  max mem: 4355
Epoch: [6]  [ 30/104]  eta: 0:01:13  lr: 0.000050  loss: 0.2539 (0.2399)  loss_classifier: 0.0629 (0.0681)  loss_box_reg: 0.1596 (0.1623)  loss_objectness: 0.0014 (0.0021)  loss_rpn_box_reg: 0.0093 (0.0074)  time: 0.9755  data: 0.0197  max mem: 4355
Epoch: [6]  [ 40/104]  eta: 0:01:03  lr: 0.000050  loss: 0.2085 (0.2329)  loss_classifier: 0.0577 (0.0660)  loss_box_reg: 0.1449 (0.1580)  loss_objectness: 0.0011 (0.0020)  loss_rpn_box_reg: 0.0043 (0.0069)  time: 0.9653  data: 0.0206  max mem: 4355
Epoch: [6]  [ 50/104]  eta: 0:00:53  lr: 0.000050  loss: 0.2085 (0.2312)  loss_classifier: 0.0580 (0.0651)  loss_box_reg: 0.1435 (0.1573)  loss_objectness: 0.0022 (0.0022)  loss_rpn_box_reg: 0.0041 (0.0067)  time: 0.9580  data: 0.0208  max mem: 4355
Epoch: [6]  [ 60/104]  eta: 0:00:42  lr: 0.000050  loss: 0.2050 (0.2286)  loss_classifier: 0.0580 (0.0648)  loss_box_reg: 0.1381 (0.1549)  loss_objectness: 0.0019 (0.0023)  loss_rpn_box_reg: 0.0043 (0.0066)  time: 0.9488  data: 0.0201  max mem: 4355
Epoch: [6]  [ 70/104]  eta: 0:00:33  lr: 0.000050  loss: 0.2039 (0.2241)  loss_classifier: 0.0560 (0.0632)  loss_box_reg: 0.1381 (0.1522)  loss_objectness: 0.0018 (0.0023)  loss_rpn_box_reg: 0.0043 (0.0064)  time: 0.9452  data: 0.0215  max mem: 4355
Epoch: [6]  [ 80/104]  eta: 0:00:23  lr: 0.000050  loss: 0.2100 (0.2229)  loss_classifier: 0.0583 (0.0633)  loss_box_reg: 0.1404 (0.1511)  loss_objectness: 0.0018 (0.0023)  loss_rpn_box_reg: 0.0044 (0.0063)  time: 0.9473  data: 0.0207  max mem: 4355
Epoch: [6]  [ 90/104]  eta: 0:00:13  lr: 0.000050  loss: 0.2112 (0.2228)  loss_classifier: 0.0583 (0.0631)  loss_box_reg: 0.1423 (0.1507)  loss_objectness: 0.0018 (0.0023)  loss_rpn_box_reg: 0.0045 (0.0067)  time: 0.9526  data: 0.0200  max mem: 4355
Epoch: [6]  [100/104]  eta: 0:00:03  lr: 0.000050  loss: 0.1688 (0.2184)  loss_classifier: 0.0461 (0.0620)  loss_box_reg: 0.1179 (0.1478)  loss_objectness: 0.0013 (0.0022)  loss_rpn_box_reg: 0.0037 (0.0065)  time: 0.9493  data: 0.0192  max mem: 4355
Epoch: [6]  [103/104]  eta: 0:00:00  lr: 0.000050  loss: 0.1812 (0.2190)  loss_classifier: 0.0516 (0.0623)  loss_box_reg: 0.1246 (0.1480)  loss_objectness: 0.0013 (0.0022)  loss_rpn_box_reg: 0.0043 (0.0065)  time: 0.9515  data: 0.0192  max mem: 4355
Epoch: [6] Total time: 0:01:40 (0.9666 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:43  model_time: 0.5862 (0.5862)  evaluator_time: 0.0531 (0.0531)  time: 1.6826  data: 1.0199  max mem: 4355
Test:  [25/26]  eta: 0:00:00  model_time: 0.4025 (0.4096)  evaluator_time: 0.0108 (0.0173)  time: 0.4406  data: 0.0175  max mem: 4355
Test: Total time: 0:00:13 (0.5091 s / it)
Averaged stats: model_time: 0.4025 (0.4096)  evaluator_time: 0.0108 (0.0173)
Accumulating evaluation results...
DONE (t=0.14s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.538
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.894
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.596
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.369
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.542
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.478
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.226
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.535
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.610
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.431
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.614
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.552
Epoch: [7]  [  0/104]  eta: 0:04:01  lr: 0.000050  loss: 0.1735 (0.1735)  loss_classifier: 0.0524 (0.0524)  loss_box_reg: 0.1117 (0.1117)  loss_objectness: 0.0007 (0.0007)  loss_rpn_box_reg: 0.0087 (0.0087)  time: 2.3236  data: 1.1759  max mem: 4355
Epoch: [7]  [ 10/104]  eta: 0:01:42  lr: 0.000050  loss: 0.2170 (0.2226)  loss_classifier: 0.0638 (0.0633)  loss_box_reg: 0.1557 (0.1511)  loss_objectness: 0.0025 (0.0021)  loss_rpn_box_reg: 0.0053 (0.0061)  time: 1.0877  data: 0.1215  max mem: 4355
Epoch: [7]  [ 20/104]  eta: 0:01:27  lr: 0.000050  loss: 0.2234 (0.2211)  loss_classifier: 0.0653 (0.0631)  loss_box_reg: 0.1588 (0.1487)  loss_objectness: 0.0023 (0.0022)  loss_rpn_box_reg: 0.0053 (0.0072)  time: 0.9759  data: 0.0198  max mem: 4355
Epoch: [7]  [ 30/104]  eta: 0:01:15  lr: 0.000050  loss: 0.2398 (0.2309)  loss_classifier: 0.0718 (0.0656)  loss_box_reg: 0.1588 (0.1551)  loss_objectness: 0.0019 (0.0024)  loss_rpn_box_reg: 0.0068 (0.0078)  time: 0.9786  data: 0.0213  max mem: 4355
Epoch: [7]  [ 40/104]  eta: 0:01:04  lr: 0.000050  loss: 0.2227 (0.2237)  loss_classifier: 0.0631 (0.0640)  loss_box_reg: 0.1458 (0.1501)  loss_objectness: 0.0015 (0.0023)  loss_rpn_box_reg: 0.0052 (0.0073)  time: 0.9627  data: 0.0193  max mem: 4355
Epoch: [7]  [ 50/104]  eta: 0:00:53  lr: 0.000050  loss: 0.2049 (0.2202)  loss_classifier: 0.0573 (0.0635)  loss_box_reg: 0.1324 (0.1476)  loss_objectness: 0.0013 (0.0022)  loss_rpn_box_reg: 0.0048 (0.0069)  time: 0.9543  data: 0.0199  max mem: 4355
Epoch: [7]  [ 60/104]  eta: 0:00:43  lr: 0.000050  loss: 0.1857 (0.2151)  loss_classifier: 0.0554 (0.0616)  loss_box_reg: 0.1299 (0.1448)  loss_objectness: 0.0016 (0.0021)  loss_rpn_box_reg: 0.0039 (0.0066)  time: 0.9568  data: 0.0227  max mem: 4355
Epoch: [7]  [ 70/104]  eta: 0:00:33  lr: 0.000050  loss: 0.1857 (0.2130)  loss_classifier: 0.0501 (0.0611)  loss_box_reg: 0.1299 (0.1434)  loss_objectness: 0.0009 (0.0020)  loss_rpn_box_reg: 0.0044 (0.0065)  time: 0.9628  data: 0.0259  max mem: 4355
Epoch: [7]  [ 80/104]  eta: 0:00:23  lr: 0.000050  loss: 0.2195 (0.2148)  loss_classifier: 0.0552 (0.0615)  loss_box_reg: 0.1474 (0.1447)  loss_objectness: 0.0015 (0.0021)  loss_rpn_box_reg: 0.0059 (0.0066)  time: 0.9571  data: 0.0234  max mem: 4355
Epoch: [7]  [ 90/104]  eta: 0:00:13  lr: 0.000050  loss: 0.2476 (0.2177)  loss_classifier: 0.0651 (0.0621)  loss_box_reg: 0.1683 (0.1469)  loss_objectness: 0.0015 (0.0021)  loss_rpn_box_reg: 0.0055 (0.0066)  time: 0.9523  data: 0.0205  max mem: 4355
Epoch: [7]  [100/104]  eta: 0:00:03  lr: 0.000050  loss: 0.2052 (0.2143)  loss_classifier: 0.0567 (0.0611)  loss_box_reg: 0.1451 (0.1447)  loss_objectness: 0.0009 (0.0020)  loss_rpn_box_reg: 0.0048 (0.0064)  time: 0.9548  data: 0.0199  max mem: 4355
Epoch: [7]  [103/104]  eta: 0:00:00  lr: 0.000050  loss: 0.2052 (0.2158)  loss_classifier: 0.0572 (0.0615)  loss_box_reg: 0.1451 (0.1459)  loss_objectness: 0.0010 (0.0020)  loss_rpn_box_reg: 0.0049 (0.0065)  time: 0.9557  data: 0.0194  max mem: 4355
Epoch: [7] Total time: 0:01:41 (0.9755 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:29  model_time: 0.4586 (0.4586)  evaluator_time: 0.0161 (0.0161)  time: 1.1438  data: 0.6522  max mem: 4355
Test:  [25/26]  eta: 0:00:00  model_time: 0.4044 (0.4062)  evaluator_time: 0.0137 (0.0190)  time: 0.4496  data: 0.0196  max mem: 4355
Test: Total time: 0:00:12 (0.4794 s / it)
Averaged stats: model_time: 0.4044 (0.4062)  evaluator_time: 0.0137 (0.0190)
Accumulating evaluation results...
DONE (t=0.07s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.539
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.891
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.596
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.370
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.541
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.484
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.220
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.530
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.605
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.431
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.601
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.577
Epoch: [8]  [  0/104]  eta: 0:02:53  lr: 0.000050  loss: 0.1753 (0.1753)  loss_classifier: 0.0522 (0.0522)  loss_box_reg: 0.1133 (0.1133)  loss_objectness: 0.0014 (0.0014)  loss_rpn_box_reg: 0.0083 (0.0083)  time: 1.6643  data: 0.5835  max mem: 4355
Epoch: [8]  [ 10/104]  eta: 0:01:37  lr: 0.000050  loss: 0.2261 (0.2278)  loss_classifier: 0.0617 (0.0615)  loss_box_reg: 0.1555 (0.1558)  loss_objectness: 0.0024 (0.0026)  loss_rpn_box_reg: 0.0065 (0.0079)  time: 1.0320  data: 0.0711  max mem: 4355
Epoch: [8]  [ 20/104]  eta: 0:01:24  lr: 0.000050  loss: 0.2177 (0.2206)  loss_classifier: 0.0589 (0.0616)  loss_box_reg: 0.1358 (0.1496)  loss_objectness: 0.0020 (0.0023)  loss_rpn_box_reg: 0.0059 (0.0071)  time: 0.9735  data: 0.0197  max mem: 4355
Epoch: [8]  [ 30/104]  eta: 0:01:13  lr: 0.000050  loss: 0.1976 (0.2221)  loss_classifier: 0.0586 (0.0619)  loss_box_reg: 0.1358 (0.1511)  loss_objectness: 0.0019 (0.0022)  loss_rpn_box_reg: 0.0055 (0.0069)  time: 0.9794  data: 0.0218  max mem: 4355
Epoch: [8]  [ 40/104]  eta: 0:01:03  lr: 0.000050  loss: 0.1850 (0.2166)  loss_classifier: 0.0518 (0.0603)  loss_box_reg: 0.1283 (0.1468)  loss_objectness: 0.0015 (0.0024)  loss_rpn_box_reg: 0.0053 (0.0072)  time: 0.9735  data: 0.0227  max mem: 4355
Epoch: [8]  [ 50/104]  eta: 0:00:52  lr: 0.000050  loss: 0.1868 (0.2197)  loss_classifier: 0.0577 (0.0623)  loss_box_reg: 0.1283 (0.1483)  loss_objectness: 0.0008 (0.0021)  loss_rpn_box_reg: 0.0043 (0.0070)  time: 0.9556  data: 0.0201  max mem: 4355
Epoch: [8]  [ 60/104]  eta: 0:00:42  lr: 0.000050  loss: 0.1878 (0.2183)  loss_classifier: 0.0593 (0.0623)  loss_box_reg: 0.1294 (0.1472)  loss_objectness: 0.0010 (0.0021)  loss_rpn_box_reg: 0.0038 (0.0067)  time: 0.9475  data: 0.0201  max mem: 4355
Epoch: [8]  [ 70/104]  eta: 0:00:33  lr: 0.000050  loss: 0.1817 (0.2130)  loss_classifier: 0.0540 (0.0611)  loss_box_reg: 0.1209 (0.1434)  loss_objectness: 0.0012 (0.0020)  loss_rpn_box_reg: 0.0035 (0.0064)  time: 0.9480  data: 0.0212  max mem: 4355
Epoch: [8]  [ 80/104]  eta: 0:00:23  lr: 0.000050  loss: 0.1902 (0.2135)  loss_classifier: 0.0536 (0.0612)  loss_box_reg: 0.1259 (0.1437)  loss_objectness: 0.0015 (0.0020)  loss_rpn_box_reg: 0.0059 (0.0067)  time: 0.9472  data: 0.0207  max mem: 4355
Epoch: [8]  [ 90/104]  eta: 0:00:13  lr: 0.000050  loss: 0.2300 (0.2139)  loss_classifier: 0.0589 (0.0611)  loss_box_reg: 0.1582 (0.1444)  loss_objectness: 0.0015 (0.0020)  loss_rpn_box_reg: 0.0067 (0.0065)  time: 0.9505  data: 0.0206  max mem: 4355
Epoch: [8]  [100/104]  eta: 0:00:03  lr: 0.000050  loss: 0.2215 (0.2139)  loss_classifier: 0.0590 (0.0610)  loss_box_reg: 0.1558 (0.1447)  loss_objectness: 0.0013 (0.0019)  loss_rpn_box_reg: 0.0047 (0.0064)  time: 0.9548  data: 0.0209  max mem: 4355
Epoch: [8]  [103/104]  eta: 0:00:00  lr: 0.000050  loss: 0.2166 (0.2152)  loss_classifier: 0.0605 (0.0613)  loss_box_reg: 0.1458 (0.1454)  loss_objectness: 0.0012 (0.0020)  loss_rpn_box_reg: 0.0048 (0.0065)  time: 0.9511  data: 0.0196  max mem: 4355
Epoch: [8] Total time: 0:01:40 (0.9669 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:32  model_time: 0.5011 (0.5011)  evaluator_time: 0.0222 (0.0222)  time: 1.2310  data: 0.6990  max mem: 4355
Test:  [25/26]  eta: 0:00:00  model_time: 0.4050 (0.4083)  evaluator_time: 0.0148 (0.0189)  time: 0.4512  data: 0.0197  max mem: 4355
Test: Total time: 0:00:12 (0.4826 s / it)
Averaged stats: model_time: 0.4050 (0.4083)  evaluator_time: 0.0148 (0.0189)
Accumulating evaluation results...
DONE (t=0.08s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.538
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.891
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.599
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.370
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.540
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.483
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.220
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.531
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.605
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.432
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.603
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.576
Epoch: [9]  [  0/104]  eta: 0:02:59  lr: 0.000005  loss: 0.2747 (0.2747)  loss_classifier: 0.0861 (0.0861)  loss_box_reg: 0.1769 (0.1769)  loss_objectness: 0.0057 (0.0057)  loss_rpn_box_reg: 0.0060 (0.0060)  time: 1.7233  data: 0.7040  max mem: 4355
Epoch: [9]  [ 10/104]  eta: 0:01:37  lr: 0.000005  loss: 0.1894 (0.1870)  loss_classifier: 0.0490 (0.0523)  loss_box_reg: 0.1344 (0.1286)  loss_objectness: 0.0014 (0.0021)  loss_rpn_box_reg: 0.0044 (0.0040)  time: 1.0415  data: 0.0802  max mem: 4355
Epoch: [9]  [ 20/104]  eta: 0:01:25  lr: 0.000005  loss: 0.1927 (0.2020)  loss_classifier: 0.0524 (0.0572)  loss_box_reg: 0.1344 (0.1372)  loss_objectness: 0.0014 (0.0024)  loss_rpn_box_reg: 0.0044 (0.0053)  time: 0.9822  data: 0.0212  max mem: 4355
Epoch: [9]  [ 30/104]  eta: 0:01:14  lr: 0.000005  loss: 0.2098 (0.2164)  loss_classifier: 0.0611 (0.0615)  loss_box_reg: 0.1394 (0.1464)  loss_objectness: 0.0016 (0.0024)  loss_rpn_box_reg: 0.0066 (0.0061)  time: 0.9863  data: 0.0233  max mem: 4355
Epoch: [9]  [ 40/104]  eta: 0:01:03  lr: 0.000005  loss: 0.1981 (0.2144)  loss_classifier: 0.0548 (0.0604)  loss_box_reg: 0.1372 (0.1452)  loss_objectness: 0.0016 (0.0024)  loss_rpn_box_reg: 0.0044 (0.0064)  time: 0.9718  data: 0.0216  max mem: 4355
Epoch: [9]  [ 50/104]  eta: 0:00:53  lr: 0.000005  loss: 0.1892 (0.2187)  loss_classifier: 0.0589 (0.0623)  loss_box_reg: 0.1184 (0.1477)  loss_objectness: 0.0016 (0.0023)  loss_rpn_box_reg: 0.0045 (0.0064)  time: 0.9592  data: 0.0218  max mem: 4355
Epoch: [9]  [ 60/104]  eta: 0:00:43  lr: 0.000005  loss: 0.2039 (0.2182)  loss_classifier: 0.0611 (0.0622)  loss_box_reg: 0.1382 (0.1475)  loss_objectness: 0.0013 (0.0022)  loss_rpn_box_reg: 0.0050 (0.0064)  time: 0.9484  data: 0.0203  max mem: 4355
Epoch: [9]  [ 70/104]  eta: 0:00:33  lr: 0.000005  loss: 0.1871 (0.2166)  loss_classifier: 0.0535 (0.0612)  loss_box_reg: 0.1327 (0.1469)  loss_objectness: 0.0010 (0.0021)  loss_rpn_box_reg: 0.0047 (0.0064)  time: 0.9433  data: 0.0204  max mem: 4355
Epoch: [9]  [ 80/104]  eta: 0:00:23  lr: 0.000005  loss: 0.1871 (0.2146)  loss_classifier: 0.0535 (0.0604)  loss_box_reg: 0.1312 (0.1457)  loss_objectness: 0.0012 (0.0021)  loss_rpn_box_reg: 0.0055 (0.0064)  time: 0.9457  data: 0.0215  max mem: 4355
Epoch: [9]  [ 90/104]  eta: 0:00:13  lr: 0.000005  loss: 0.2013 (0.2137)  loss_classifier: 0.0550 (0.0605)  loss_box_reg: 0.1380 (0.1448)  loss_objectness: 0.0015 (0.0020)  loss_rpn_box_reg: 0.0062 (0.0064)  time: 0.9451  data: 0.0196  max mem: 4355
Epoch: [9]  [100/104]  eta: 0:00:03  lr: 0.000005  loss: 0.2118 (0.2138)  loss_classifier: 0.0558 (0.0606)  loss_box_reg: 0.1422 (0.1448)  loss_objectness: 0.0015 (0.0020)  loss_rpn_box_reg: 0.0062 (0.0064)  time: 0.9498  data: 0.0186  max mem: 4355
Epoch: [9]  [103/104]  eta: 0:00:00  lr: 0.000005  loss: 0.1894 (0.2135)  loss_classifier: 0.0547 (0.0606)  loss_box_reg: 0.1295 (0.1444)  loss_objectness: 0.0014 (0.0020)  loss_rpn_box_reg: 0.0052 (0.0064)  time: 0.9504  data: 0.0183  max mem: 4355
Epoch: [9] Total time: 0:01:40 (0.9676 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:27  model_time: 0.4471 (0.4471)  evaluator_time: 0.0167 (0.0167)  time: 1.0498  data: 0.5740  max mem: 4355
Test:  [25/26]  eta: 0:00:00  model_time: 0.4034 (0.4060)  evaluator_time: 0.0125 (0.0186)  time: 0.4467  data: 0.0193  max mem: 4355
Test: Total time: 0:00:12 (0.4761 s / it)
Averaged stats: model_time: 0.4034 (0.4060)  evaluator_time: 0.0125 (0.0186)
Accumulating evaluation results...
DONE (t=0.07s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.538
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.891
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.599
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.370
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.540
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.483
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.220
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.530
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.605
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.431
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.602
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.576
Epoch: [10]  [  0/104]  eta: 0:03:00  lr: 0.000005  loss: 0.2670 (0.2670)  loss_classifier: 0.0906 (0.0906)  loss_box_reg: 0.1686 (0.1686)  loss_objectness: 0.0016 (0.0016)  loss_rpn_box_reg: 0.0063 (0.0063)  time: 1.7310  data: 0.7506  max mem: 4355
Epoch: [10]  [ 10/104]  eta: 0:01:37  lr: 0.000005  loss: 0.2108 (0.2081)  loss_classifier: 0.0610 (0.0627)  loss_box_reg: 0.1423 (0.1377)  loss_objectness: 0.0014 (0.0018)  loss_rpn_box_reg: 0.0062 (0.0059)  time: 1.0400  data: 0.0838  max mem: 4355
Epoch: [10]  [ 20/104]  eta: 0:01:25  lr: 0.000005  loss: 0.2108 (0.2098)  loss_classifier: 0.0595 (0.0606)  loss_box_reg: 0.1423 (0.1416)  loss_objectness: 0.0012 (0.0016)  loss_rpn_box_reg: 0.0058 (0.0061)  time: 0.9781  data: 0.0190  max mem: 4355
Epoch: [10]  [ 30/104]  eta: 0:01:14  lr: 0.000005  loss: 0.2136 (0.2086)  loss_classifier: 0.0644 (0.0600)  loss_box_reg: 0.1432 (0.1405)  loss_objectness: 0.0014 (0.0018)  loss_rpn_box_reg: 0.0056 (0.0063)  time: 0.9801  data: 0.0208  max mem: 4355
Epoch: [10]  [ 40/104]  eta: 0:01:03  lr: 0.000005  loss: 0.1992 (0.2072)  loss_classifier: 0.0568 (0.0598)  loss_box_reg: 0.1350 (0.1397)  loss_objectness: 0.0013 (0.0017)  loss_rpn_box_reg: 0.0039 (0.0059)  time: 0.9665  data: 0.0199  max mem: 4355
Epoch: [10]  [ 50/104]  eta: 0:00:53  lr: 0.000005  loss: 0.2019 (0.2115)  loss_classifier: 0.0568 (0.0614)  loss_box_reg: 0.1417 (0.1420)  loss_objectness: 0.0010 (0.0017)  loss_rpn_box_reg: 0.0048 (0.0063)  time: 0.9574  data: 0.0201  max mem: 4355
Epoch: [10]  [ 60/104]  eta: 0:00:43  lr: 0.000005  loss: 0.2197 (0.2124)  loss_classifier: 0.0627 (0.0620)  loss_box_reg: 0.1424 (0.1422)  loss_objectness: 0.0014 (0.0018)  loss_rpn_box_reg: 0.0053 (0.0064)  time: 0.9508  data: 0.0204  max mem: 4355
Epoch: [10]  [ 70/104]  eta: 0:00:33  lr: 0.000005  loss: 0.2106 (0.2114)  loss_classifier: 0.0603 (0.0614)  loss_box_reg: 0.1370 (0.1418)  loss_objectness: 0.0018 (0.0019)  loss_rpn_box_reg: 0.0050 (0.0063)  time: 0.9439  data: 0.0194  max mem: 4355
Epoch: [10]  [ 80/104]  eta: 0:00:23  lr: 0.000005  loss: 0.1949 (0.2139)  loss_classifier: 0.0611 (0.0620)  loss_box_reg: 0.1414 (0.1436)  loss_objectness: 0.0018 (0.0019)  loss_rpn_box_reg: 0.0056 (0.0063)  time: 0.9463  data: 0.0203  max mem: 4355
Epoch: [10]  [ 90/104]  eta: 0:00:13  lr: 0.000005  loss: 0.2320 (0.2142)  loss_classifier: 0.0630 (0.0619)  loss_box_reg: 0.1552 (0.1440)  loss_objectness: 0.0010 (0.0019)  loss_rpn_box_reg: 0.0054 (0.0065)  time: 0.9535  data: 0.0220  max mem: 4355
Epoch: [10]  [100/104]  eta: 0:00:03  lr: 0.000005  loss: 0.2157 (0.2144)  loss_classifier: 0.0597 (0.0614)  loss_box_reg: 0.1539 (0.1447)  loss_objectness: 0.0017 (0.0019)  loss_rpn_box_reg: 0.0049 (0.0064)  time: 0.9513  data: 0.0200  max mem: 4355
Epoch: [10]  [103/104]  eta: 0:00:00  lr: 0.000005  loss: 0.1912 (0.2138)  loss_classifier: 0.0562 (0.0614)  loss_box_reg: 0.1265 (0.1441)  loss_objectness: 0.0012 (0.0019)  loss_rpn_box_reg: 0.0047 (0.0064)  time: 0.9544  data: 0.0200  max mem: 4355
Epoch: [10] Total time: 0:01:40 (0.9676 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:51  model_time: 0.5838 (0.5838)  evaluator_time: 0.0634 (0.0634)  time: 1.9818  data: 1.3005  max mem: 4355
Test:  [25/26]  eta: 0:00:00  model_time: 0.4025 (0.4101)  evaluator_time: 0.0123 (0.0182)  time: 0.4417  data: 0.0180  max mem: 4355
Test: Total time: 0:00:13 (0.5079 s / it)
Averaged stats: model_time: 0.4025 (0.4101)  evaluator_time: 0.0123 (0.0182)
Accumulating evaluation results...
DONE (t=0.14s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.537
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.891
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.599
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.370
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.540
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.483
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.220
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.530
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.605
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.431
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.602
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.576
Epoch: [11]  [  0/104]  eta: 0:03:53  lr: 0.000005  loss: 0.1864 (0.1864)  loss_classifier: 0.0484 (0.0484)  loss_box_reg: 0.1316 (0.1316)  loss_objectness: 0.0012 (0.0012)  loss_rpn_box_reg: 0.0052 (0.0052)  time: 2.2469  data: 1.0963  max mem: 4355
Epoch: [11]  [ 10/104]  eta: 0:01:41  lr: 0.000005  loss: 0.1747 (0.1639)  loss_classifier: 0.0476 (0.0478)  loss_box_reg: 0.1178 (0.1094)  loss_objectness: 0.0012 (0.0023)  loss_rpn_box_reg: 0.0047 (0.0044)  time: 1.0772  data: 0.1148  max mem: 4355
Epoch: [11]  [ 20/104]  eta: 0:01:26  lr: 0.000005  loss: 0.1747 (0.1792)  loss_classifier: 0.0476 (0.0501)  loss_box_reg: 0.1197 (0.1222)  loss_objectness: 0.0013 (0.0022)  loss_rpn_box_reg: 0.0045 (0.0047)  time: 0.9732  data: 0.0179  max mem: 4355
Epoch: [11]  [ 30/104]  eta: 0:01:15  lr: 0.000005  loss: 0.1969 (0.1938)  loss_classifier: 0.0587 (0.0544)  loss_box_reg: 0.1443 (0.1322)  loss_objectness: 0.0014 (0.0021)  loss_rpn_box_reg: 0.0052 (0.0051)  time: 0.9813  data: 0.0203  max mem: 4355
Epoch: [11]  [ 40/104]  eta: 0:01:04  lr: 0.000005  loss: 0.2063 (0.1954)  loss_classifier: 0.0594 (0.0556)  loss_box_reg: 0.1454 (0.1324)  loss_objectness: 0.0016 (0.0022)  loss_rpn_box_reg: 0.0053 (0.0052)  time: 0.9670  data: 0.0202  max mem: 4355
Epoch: [11]  [ 50/104]  eta: 0:00:53  lr: 0.000005  loss: 0.2383 (0.2033)  loss_classifier: 0.0673 (0.0578)  loss_box_reg: 0.1593 (0.1375)  loss_objectness: 0.0021 (0.0023)  loss_rpn_box_reg: 0.0058 (0.0057)  time: 0.9729  data: 0.0242  max mem: 4355
Epoch: [11]  [ 60/104]  eta: 0:00:43  lr: 0.000005  loss: 0.2416 (0.2062)  loss_classifier: 0.0650 (0.0582)  loss_box_reg: 0.1623 (0.1399)  loss_objectness: 0.0024 (0.0023)  loss_rpn_box_reg: 0.0072 (0.0058)  time: 0.9699  data: 0.0266  max mem: 4355
Epoch: [11]  [ 70/104]  eta: 0:00:33  lr: 0.000005  loss: 0.2074 (0.2100)  loss_classifier: 0.0581 (0.0596)  loss_box_reg: 0.1457 (0.1421)  loss_objectness: 0.0015 (0.0022)  loss_rpn_box_reg: 0.0064 (0.0061)  time: 0.9442  data: 0.0215  max mem: 4355
Epoch: [11]  [ 80/104]  eta: 0:00:23  lr: 0.000005  loss: 0.2106 (0.2102)  loss_classifier: 0.0583 (0.0594)  loss_box_reg: 0.1457 (0.1422)  loss_objectness: 0.0015 (0.0022)  loss_rpn_box_reg: 0.0058 (0.0063)  time: 0.9448  data: 0.0211  max mem: 4355
Epoch: [11]  [ 90/104]  eta: 0:00:13  lr: 0.000005  loss: 0.2203 (0.2131)  loss_classifier: 0.0692 (0.0608)  loss_box_reg: 0.1496 (0.1436)  loss_objectness: 0.0014 (0.0023)  loss_rpn_box_reg: 0.0054 (0.0064)  time: 0.9538  data: 0.0220  max mem: 4355
Epoch: [11]  [100/104]  eta: 0:00:03  lr: 0.000005  loss: 0.2337 (0.2140)  loss_classifier: 0.0699 (0.0611)  loss_box_reg: 0.1496 (0.1442)  loss_objectness: 0.0014 (0.0022)  loss_rpn_box_reg: 0.0060 (0.0064)  time: 0.9520  data: 0.0196  max mem: 4355
Epoch: [11]  [103/104]  eta: 0:00:00  lr: 0.000005  loss: 0.2203 (0.2137)  loss_classifier: 0.0644 (0.0612)  loss_box_reg: 0.1460 (0.1439)  loss_objectness: 0.0014 (0.0022)  loss_rpn_box_reg: 0.0056 (0.0064)  time: 0.9556  data: 0.0197  max mem: 4355
Epoch: [11] Total time: 0:01:41 (0.9749 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:34  model_time: 0.4899 (0.4899)  evaluator_time: 0.0167 (0.0167)  time: 1.3324  data: 0.8204  max mem: 4355
Test:  [25/26]  eta: 0:00:00  model_time: 0.4024 (0.4067)  evaluator_time: 0.0100 (0.0185)  time: 0.4487  data: 0.0199  max mem: 4355
Test: Total time: 0:00:12 (0.4844 s / it)
Averaged stats: model_time: 0.4024 (0.4067)  evaluator_time: 0.0100 (0.0185)
Accumulating evaluation results...
DONE (t=0.14s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.538
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.891
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.599
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.370
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.540
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.484
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.220
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.530
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.605
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.431
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.602
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.577
Epoch: [12]  [  0/104]  eta: 0:03:13  lr: 0.000001  loss: 0.3097 (0.3097)  loss_classifier: 0.0874 (0.0874)  loss_box_reg: 0.2138 (0.2138)  loss_objectness: 0.0014 (0.0014)  loss_rpn_box_reg: 0.0070 (0.0070)  time: 1.8569  data: 0.8821  max mem: 4355
Epoch: [12]  [ 10/104]  eta: 0:01:38  lr: 0.000001  loss: 0.2367 (0.2286)  loss_classifier: 0.0637 (0.0666)  loss_box_reg: 0.1536 (0.1514)  loss_objectness: 0.0015 (0.0026)  loss_rpn_box_reg: 0.0074 (0.0080)  time: 1.0436  data: 0.0948  max mem: 4355
Epoch: [12]  [ 20/104]  eta: 0:01:25  lr: 0.000001  loss: 0.2173 (0.2200)  loss_classifier: 0.0605 (0.0612)  loss_box_reg: 0.1486 (0.1487)  loss_objectness: 0.0012 (0.0021)  loss_rpn_box_reg: 0.0070 (0.0080)  time: 0.9735  data: 0.0185  max mem: 4355
Epoch: [12]  [ 30/104]  eta: 0:01:14  lr: 0.000001  loss: 0.2139 (0.2248)  loss_classifier: 0.0637 (0.0638)  loss_box_reg: 0.1435 (0.1506)  loss_objectness: 0.0015 (0.0025)  loss_rpn_box_reg: 0.0047 (0.0078)  time: 0.9810  data: 0.0220  max mem: 4355
Epoch: [12]  [ 40/104]  eta: 0:01:03  lr: 0.000001  loss: 0.2034 (0.2156)  loss_classifier: 0.0597 (0.0609)  loss_box_reg: 0.1373 (0.1452)  loss_objectness: 0.0020 (0.0023)  loss_rpn_box_reg: 0.0046 (0.0072)  time: 0.9726  data: 0.0230  max mem: 4355
Epoch: [12]  [ 50/104]  eta: 0:00:53  lr: 0.000001  loss: 0.1979 (0.2145)  loss_classifier: 0.0558 (0.0613)  loss_box_reg: 0.1276 (0.1441)  loss_objectness: 0.0014 (0.0023)  loss_rpn_box_reg: 0.0047 (0.0069)  time: 0.9568  data: 0.0208  max mem: 4355
Epoch: [12]  [ 60/104]  eta: 0:00:43  lr: 0.000001  loss: 0.2116 (0.2162)  loss_classifier: 0.0632 (0.0620)  loss_box_reg: 0.1338 (0.1453)  loss_objectness: 0.0019 (0.0023)  loss_rpn_box_reg: 0.0047 (0.0067)  time: 0.9486  data: 0.0195  max mem: 4355
Epoch: [12]  [ 70/104]  eta: 0:00:33  lr: 0.000001  loss: 0.2140 (0.2137)  loss_classifier: 0.0632 (0.0611)  loss_box_reg: 0.1433 (0.1439)  loss_objectness: 0.0019 (0.0022)  loss_rpn_box_reg: 0.0047 (0.0065)  time: 0.9485  data: 0.0200  max mem: 4355
Epoch: [12]  [ 80/104]  eta: 0:00:23  lr: 0.000001  loss: 0.2154 (0.2122)  loss_classifier: 0.0562 (0.0606)  loss_box_reg: 0.1433 (0.1430)  loss_objectness: 0.0020 (0.0022)  loss_rpn_box_reg: 0.0053 (0.0064)  time: 0.9514  data: 0.0216  max mem: 4355
Epoch: [12]  [ 90/104]  eta: 0:00:13  lr: 0.000001  loss: 0.1897 (0.2136)  loss_classifier: 0.0550 (0.0612)  loss_box_reg: 0.1214 (0.1437)  loss_objectness: 0.0020 (0.0022)  loss_rpn_box_reg: 0.0053 (0.0065)  time: 0.9539  data: 0.0213  max mem: 4355
Epoch: [12]  [100/104]  eta: 0:00:03  lr: 0.000001  loss: 0.1897 (0.2126)  loss_classifier: 0.0550 (0.0609)  loss_box_reg: 0.1214 (0.1431)  loss_objectness: 0.0023 (0.0022)  loss_rpn_box_reg: 0.0047 (0.0064)  time: 0.9551  data: 0.0203  max mem: 4355
Epoch: [12]  [103/104]  eta: 0:00:00  lr: 0.000001  loss: 0.2109 (0.2136)  loss_classifier: 0.0550 (0.0610)  loss_box_reg: 0.1340 (0.1440)  loss_objectness: 0.0023 (0.0022)  loss_rpn_box_reg: 0.0045 (0.0064)  time: 0.9519  data: 0.0196  max mem: 4355
Epoch: [12] Total time: 0:01:40 (0.9695 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:29  model_time: 0.4803 (0.4803)  evaluator_time: 0.0165 (0.0165)  time: 1.1303  data: 0.6133  max mem: 4355
Test:  [25/26]  eta: 0:00:00  model_time: 0.4036 (0.4066)  evaluator_time: 0.0138 (0.0333)  time: 0.4699  data: 0.0200  max mem: 4355
Test: Total time: 0:00:12 (0.4934 s / it)
Averaged stats: model_time: 0.4036 (0.4066)  evaluator_time: 0.0138 (0.0333)
Accumulating evaluation results...
DONE (t=0.08s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.538
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.891
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.599
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.370
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.540
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.484
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.220
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.530
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.605
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.431
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.602
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.577
Epoch: [13]  [  0/104]  eta: 0:02:59  lr: 0.000001  loss: 0.3098 (0.3098)  loss_classifier: 0.0753 (0.0753)  loss_box_reg: 0.2196 (0.2196)  loss_objectness: 0.0041 (0.0041)  loss_rpn_box_reg: 0.0108 (0.0108)  time: 1.7234  data: 0.7536  max mem: 4355
Epoch: [13]  [ 10/104]  eta: 0:01:37  lr: 0.000001  loss: 0.2109 (0.2131)  loss_classifier: 0.0570 (0.0617)  loss_box_reg: 0.1349 (0.1434)  loss_objectness: 0.0015 (0.0020)  loss_rpn_box_reg: 0.0058 (0.0061)  time: 1.0362  data: 0.0849  max mem: 4355
Epoch: [13]  [ 20/104]  eta: 0:01:24  lr: 0.000001  loss: 0.1885 (0.2113)  loss_classifier: 0.0562 (0.0616)  loss_box_reg: 0.1265 (0.1410)  loss_objectness: 0.0013 (0.0021)  loss_rpn_box_reg: 0.0049 (0.0066)  time: 0.9726  data: 0.0186  max mem: 4355
Epoch: [13]  [ 30/104]  eta: 0:01:13  lr: 0.000001  loss: 0.1926 (0.2183)  loss_classifier: 0.0510 (0.0636)  loss_box_reg: 0.1303 (0.1458)  loss_objectness: 0.0018 (0.0021)  loss_rpn_box_reg: 0.0049 (0.0068)  time: 0.9742  data: 0.0205  max mem: 4355
Epoch: [13]  [ 40/104]  eta: 0:01:03  lr: 0.000001  loss: 0.2001 (0.2155)  loss_classifier: 0.0581 (0.0621)  loss_box_reg: 0.1363 (0.1443)  loss_objectness: 0.0018 (0.0022)  loss_rpn_box_reg: 0.0054 (0.0069)  time: 0.9651  data: 0.0220  max mem: 4355
Epoch: [13]  [ 50/104]  eta: 0:00:52  lr: 0.000001  loss: 0.1935 (0.2104)  loss_classifier: 0.0557 (0.0600)  loss_box_reg: 0.1278 (0.1416)  loss_objectness: 0.0015 (0.0020)  loss_rpn_box_reg: 0.0054 (0.0068)  time: 0.9567  data: 0.0220  max mem: 4355
Epoch: [13]  [ 60/104]  eta: 0:00:42  lr: 0.000001  loss: 0.1718 (0.2059)  loss_classifier: 0.0490 (0.0592)  loss_box_reg: 0.1156 (0.1384)  loss_objectness: 0.0008 (0.0018)  loss_rpn_box_reg: 0.0043 (0.0065)  time: 0.9447  data: 0.0201  max mem: 4355
Epoch: [13]  [ 70/104]  eta: 0:00:32  lr: 0.000001  loss: 0.1823 (0.2090)  loss_classifier: 0.0546 (0.0602)  loss_box_reg: 0.1268 (0.1405)  loss_objectness: 0.0011 (0.0018)  loss_rpn_box_reg: 0.0048 (0.0064)  time: 0.9419  data: 0.0202  max mem: 4355
Epoch: [13]  [ 80/104]  eta: 0:00:23  lr: 0.000001  loss: 0.2312 (0.2116)  loss_classifier: 0.0644 (0.0608)  loss_box_reg: 0.1585 (0.1424)  loss_objectness: 0.0016 (0.0018)  loss_rpn_box_reg: 0.0061 (0.0067)  time: 0.9495  data: 0.0210  max mem: 4355
Epoch: [13]  [ 90/104]  eta: 0:00:13  lr: 0.000001  loss: 0.2416 (0.2127)  loss_classifier: 0.0693 (0.0615)  loss_box_reg: 0.1554 (0.1429)  loss_objectness: 0.0014 (0.0018)  loss_rpn_box_reg: 0.0044 (0.0065)  time: 0.9490  data: 0.0198  max mem: 4355
Epoch: [13]  [100/104]  eta: 0:00:03  lr: 0.000001  loss: 0.2387 (0.2133)  loss_classifier: 0.0700 (0.0615)  loss_box_reg: 0.1548 (0.1435)  loss_objectness: 0.0011 (0.0018)  loss_rpn_box_reg: 0.0044 (0.0065)  time: 0.9502  data: 0.0191  max mem: 4355
Epoch: [13]  [103/104]  eta: 0:00:00  lr: 0.000001  loss: 0.2383 (0.2135)  loss_classifier: 0.0692 (0.0615)  loss_box_reg: 0.1554 (0.1438)  loss_objectness: 0.0011 (0.0018)  loss_rpn_box_reg: 0.0058 (0.0064)  time: 0.9506  data: 0.0187  max mem: 4355
Epoch: [13] Total time: 0:01:40 (0.9644 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:31  model_time: 0.4894 (0.4894)  evaluator_time: 0.0180 (0.0180)  time: 1.1968  data: 0.6822  max mem: 4355
Test:  [25/26]  eta: 0:00:00  model_time: 0.4034 (0.4081)  evaluator_time: 0.0100 (0.0187)  time: 0.4459  data: 0.0193  max mem: 4355
Test: Total time: 0:00:12 (0.4806 s / it)
Averaged stats: model_time: 0.4034 (0.4081)  evaluator_time: 0.0100 (0.0187)
Accumulating evaluation results...
DONE (t=0.08s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.538
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.891
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.599
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.370
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.540
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.484
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.220
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.530
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.605
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.431
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.602
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.577
Epoch: [14]  [  0/104]  eta: 0:02:54  lr: 0.000001  loss: 0.2427 (0.2427)  loss_classifier: 0.0751 (0.0751)  loss_box_reg: 0.1627 (0.1627)  loss_objectness: 0.0006 (0.0006)  loss_rpn_box_reg: 0.0043 (0.0043)  time: 1.6752  data: 0.6798  max mem: 4355
Epoch: [14]  [ 10/104]  eta: 0:01:36  lr: 0.000001  loss: 0.2261 (0.2072)  loss_classifier: 0.0647 (0.0572)  loss_box_reg: 0.1519 (0.1405)  loss_objectness: 0.0008 (0.0024)  loss_rpn_box_reg: 0.0048 (0.0071)  time: 1.0307  data: 0.0786  max mem: 4355
Epoch: [14]  [ 20/104]  eta: 0:01:24  lr: 0.000001  loss: 0.2235 (0.2199)  loss_classifier: 0.0629 (0.0620)  loss_box_reg: 0.1528 (0.1485)  loss_objectness: 0.0008 (0.0022)  loss_rpn_box_reg: 0.0051 (0.0071)  time: 0.9756  data: 0.0211  max mem: 4355
Epoch: [14]  [ 30/104]  eta: 0:01:13  lr: 0.000001  loss: 0.2235 (0.2202)  loss_classifier: 0.0585 (0.0622)  loss_box_reg: 0.1496 (0.1486)  loss_objectness: 0.0018 (0.0025)  loss_rpn_box_reg: 0.0043 (0.0069)  time: 0.9759  data: 0.0210  max mem: 4355
Epoch: [14]  [ 40/104]  eta: 0:01:03  lr: 0.000001  loss: 0.1953 (0.2185)  loss_classifier: 0.0582 (0.0618)  loss_box_reg: 0.1366 (0.1468)  loss_objectness: 0.0026 (0.0028)  loss_rpn_box_reg: 0.0045 (0.0071)  time: 0.9610  data: 0.0186  max mem: 4355
Epoch: [14]  [ 50/104]  eta: 0:00:52  lr: 0.000001  loss: 0.1914 (0.2175)  loss_classifier: 0.0642 (0.0619)  loss_box_reg: 0.1367 (0.1458)  loss_objectness: 0.0026 (0.0028)  loss_rpn_box_reg: 0.0050 (0.0070)  time: 0.9588  data: 0.0213  max mem: 4355
Epoch: [14]  [ 60/104]  eta: 0:00:42  lr: 0.000001  loss: 0.1854 (0.2098)  loss_classifier: 0.0529 (0.0600)  loss_box_reg: 0.1270 (0.1409)  loss_objectness: 0.0008 (0.0025)  loss_rpn_box_reg: 0.0043 (0.0065)  time: 0.9557  data: 0.0232  max mem: 4355
Epoch: [14]  [ 70/104]  eta: 0:00:32  lr: 0.000001  loss: 0.1882 (0.2125)  loss_classifier: 0.0543 (0.0605)  loss_box_reg: 0.1279 (0.1429)  loss_objectness: 0.0011 (0.0025)  loss_rpn_box_reg: 0.0038 (0.0066)  time: 0.9438  data: 0.0207  max mem: 4355
Epoch: [14]  [ 80/104]  eta: 0:00:23  lr: 0.000001  loss: 0.2475 (0.2164)  loss_classifier: 0.0678 (0.0615)  loss_box_reg: 0.1601 (0.1456)  loss_objectness: 0.0017 (0.0025)  loss_rpn_box_reg: 0.0057 (0.0067)  time: 0.9433  data: 0.0192  max mem: 4355
Epoch: [14]  [ 90/104]  eta: 0:00:13  lr: 0.000001  loss: 0.2331 (0.2157)  loss_classifier: 0.0657 (0.0614)  loss_box_reg: 0.1549 (0.1452)  loss_objectness: 0.0014 (0.0025)  loss_rpn_box_reg: 0.0056 (0.0066)  time: 0.9524  data: 0.0222  max mem: 4355
Epoch: [14]  [100/104]  eta: 0:00:03  lr: 0.000001  loss: 0.1883 (0.2135)  loss_classifier: 0.0579 (0.0609)  loss_box_reg: 0.1299 (0.1437)  loss_objectness: 0.0015 (0.0024)  loss_rpn_box_reg: 0.0050 (0.0064)  time: 0.9505  data: 0.0214  max mem: 4355
Epoch: [14]  [103/104]  eta: 0:00:00  lr: 0.000001  loss: 0.1883 (0.2133)  loss_classifier: 0.0579 (0.0606)  loss_box_reg: 0.1299 (0.1439)  loss_objectness: 0.0012 (0.0024)  loss_rpn_box_reg: 0.0050 (0.0064)  time: 0.9526  data: 0.0214  max mem: 4355
Epoch: [14] Total time: 0:01:40 (0.9658 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:40  model_time: 0.6157 (0.6157)  evaluator_time: 0.0845 (0.0845)  time: 1.5539  data: 0.8393  max mem: 4355
Test:  [25/26]  eta: 0:00:00  model_time: 0.4014 (0.4104)  evaluator_time: 0.0100 (0.0185)  time: 0.4420  data: 0.0185  max mem: 4355
Test: Total time: 0:00:12 (0.4922 s / it)
Averaged stats: model_time: 0.4014 (0.4104)  evaluator_time: 0.0100 (0.0185)
Accumulating evaluation results...
DONE (t=0.12s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.538
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.891
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.599
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.370
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.540
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.484
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.220
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.530
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.605
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.431
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.602
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.577
Epoch: [15]  [  0/104]  eta: 0:03:45  lr: 0.000000  loss: 0.3076 (0.3076)  loss_classifier: 0.0903 (0.0903)  loss_box_reg: 0.2101 (0.2101)  loss_objectness: 0.0010 (0.0010)  loss_rpn_box_reg: 0.0063 (0.0063)  time: 2.1664  data: 1.0755  max mem: 4355
Epoch: [15]  [ 10/104]  eta: 0:01:40  lr: 0.000000  loss: 0.2037 (0.2317)  loss_classifier: 0.0603 (0.0686)  loss_box_reg: 0.1365 (0.1550)  loss_objectness: 0.0018 (0.0020)  loss_rpn_box_reg: 0.0063 (0.0060)  time: 1.0684  data: 0.1131  max mem: 4355
Epoch: [15]  [ 20/104]  eta: 0:01:26  lr: 0.000000  loss: 0.2037 (0.2205)  loss_classifier: 0.0603 (0.0652)  loss_box_reg: 0.1371 (0.1478)  loss_objectness: 0.0017 (0.0019)  loss_rpn_box_reg: 0.0051 (0.0056)  time: 0.9718  data: 0.0202  max mem: 4355
Epoch: [15]  [ 30/104]  eta: 0:01:15  lr: 0.000000  loss: 0.2189 (0.2210)  loss_classifier: 0.0627 (0.0646)  loss_box_reg: 0.1433 (0.1484)  loss_objectness: 0.0015 (0.0021)  loss_rpn_box_reg: 0.0051 (0.0058)  time: 0.9902  data: 0.0238  max mem: 4355
Epoch: [15]  [ 40/104]  eta: 0:01:04  lr: 0.000000  loss: 0.2099 (0.2180)  loss_classifier: 0.0578 (0.0639)  loss_box_reg: 0.1419 (0.1461)  loss_objectness: 0.0022 (0.0022)  loss_rpn_box_reg: 0.0054 (0.0058)  time: 0.9815  data: 0.0229  max mem: 4355
Epoch: [15]  [ 50/104]  eta: 0:00:53  lr: 0.000000  loss: 0.2060 (0.2146)  loss_classifier: 0.0539 (0.0623)  loss_box_reg: 0.1430 (0.1444)  loss_objectness: 0.0013 (0.0021)  loss_rpn_box_reg: 0.0045 (0.0059)  time: 0.9626  data: 0.0215  max mem: 4355
Epoch: [15]  [ 60/104]  eta: 0:00:43  lr: 0.000000  loss: 0.2060 (0.2138)  loss_classifier: 0.0557 (0.0615)  loss_box_reg: 0.1471 (0.1443)  loss_objectness: 0.0013 (0.0021)  loss_rpn_box_reg: 0.0052 (0.0060)  time: 0.9545  data: 0.0226  max mem: 4355
Epoch: [15]  [ 70/104]  eta: 0:00:33  lr: 0.000000  loss: 0.2034 (0.2134)  loss_classifier: 0.0616 (0.0615)  loss_box_reg: 0.1341 (0.1440)  loss_objectness: 0.0016 (0.0020)  loss_rpn_box_reg: 0.0052 (0.0059)  time: 0.9466  data: 0.0219  max mem: 4355
Epoch: [15]  [ 80/104]  eta: 0:00:23  lr: 0.000000  loss: 0.2017 (0.2127)  loss_classifier: 0.0611 (0.0612)  loss_box_reg: 0.1341 (0.1435)  loss_objectness: 0.0018 (0.0021)  loss_rpn_box_reg: 0.0052 (0.0060)  time: 0.9463  data: 0.0213  max mem: 4355
Epoch: [15]  [ 90/104]  eta: 0:00:13  lr: 0.000000  loss: 0.2017 (0.2118)  loss_classifier: 0.0522 (0.0607)  loss_box_reg: 0.1420 (0.1430)  loss_objectness: 0.0021 (0.0021)  loss_rpn_box_reg: 0.0056 (0.0060)  time: 0.9546  data: 0.0213  max mem: 4355
Epoch: [15]  [100/104]  eta: 0:00:03  lr: 0.000000  loss: 0.2206 (0.2151)  loss_classifier: 0.0644 (0.0614)  loss_box_reg: 0.1520 (0.1450)  loss_objectness: 0.0020 (0.0021)  loss_rpn_box_reg: 0.0056 (0.0065)  time: 0.9571  data: 0.0198  max mem: 4355
Epoch: [15]  [103/104]  eta: 0:00:00  lr: 0.000000  loss: 0.2225 (0.2134)  loss_classifier: 0.0644 (0.0609)  loss_box_reg: 0.1626 (0.1440)  loss_objectness: 0.0018 (0.0021)  loss_rpn_box_reg: 0.0049 (0.0064)  time: 0.9587  data: 0.0196  max mem: 4355
Epoch: [15] Total time: 0:01:41 (0.9747 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:31  model_time: 0.4641 (0.4641)  evaluator_time: 0.0199 (0.0199)  time: 1.2298  data: 0.7273  max mem: 4355
Test:  [25/26]  eta: 0:00:00  model_time: 0.4043 (0.4066)  evaluator_time: 0.0133 (0.0185)  time: 0.4477  data: 0.0186  max mem: 4355
Test: Total time: 0:00:12 (0.4800 s / it)
Averaged stats: model_time: 0.4043 (0.4066)  evaluator_time: 0.0133 (0.0185)
Accumulating evaluation results...
DONE (t=0.07s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.538
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.891
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.599
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.370
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.540
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.484
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.220
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.530
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.605
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.431
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.602
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.577
Epoch: [16]  [  0/104]  eta: 0:02:51  lr: 0.000000  loss: 0.0866 (0.0866)  loss_classifier: 0.0232 (0.0232)  loss_box_reg: 0.0614 (0.0614)  loss_objectness: 0.0005 (0.0005)  loss_rpn_box_reg: 0.0015 (0.0015)  time: 1.6520  data: 0.6281  max mem: 4355
Epoch: [16]  [ 10/104]  eta: 0:01:37  lr: 0.000000  loss: 0.1753 (0.1950)  loss_classifier: 0.0504 (0.0575)  loss_box_reg: 0.1141 (0.1307)  loss_objectness: 0.0006 (0.0015)  loss_rpn_box_reg: 0.0042 (0.0054)  time: 1.0382  data: 0.0734  max mem: 4355
Epoch: [16]  [ 20/104]  eta: 0:01:24  lr: 0.000000  loss: 0.1802 (0.2053)  loss_classifier: 0.0531 (0.0599)  loss_box_reg: 0.1238 (0.1377)  loss_objectness: 0.0008 (0.0019)  loss_rpn_box_reg: 0.0042 (0.0057)  time: 0.9785  data: 0.0188  max mem: 4355
Epoch: [16]  [ 30/104]  eta: 0:01:13  lr: 0.000000  loss: 0.1734 (0.1981)  loss_classifier: 0.0502 (0.0569)  loss_box_reg: 0.1219 (0.1338)  loss_objectness: 0.0013 (0.0017)  loss_rpn_box_reg: 0.0040 (0.0057)  time: 0.9776  data: 0.0202  max mem: 4355
Epoch: [16]  [ 40/104]  eta: 0:01:03  lr: 0.000000  loss: 0.1840 (0.2058)  loss_classifier: 0.0507 (0.0588)  loss_box_reg: 0.1243 (0.1390)  loss_objectness: 0.0009 (0.0017)  loss_rpn_box_reg: 0.0039 (0.0064)  time: 0.9678  data: 0.0204  max mem: 4355
Epoch: [16]  [ 50/104]  eta: 0:00:52  lr: 0.000000  loss: 0.1912 (0.2074)  loss_classifier: 0.0547 (0.0595)  loss_box_reg: 0.1329 (0.1397)  loss_objectness: 0.0011 (0.0017)  loss_rpn_box_reg: 0.0049 (0.0065)  time: 0.9503  data: 0.0191  max mem: 4355
Epoch: [16]  [ 60/104]  eta: 0:00:42  lr: 0.000000  loss: 0.1912 (0.2080)  loss_classifier: 0.0547 (0.0593)  loss_box_reg: 0.1346 (0.1405)  loss_objectness: 0.0015 (0.0019)  loss_rpn_box_reg: 0.0060 (0.0064)  time: 0.9435  data: 0.0195  max mem: 4355
Epoch: [16]  [ 70/104]  eta: 0:00:32  lr: 0.000000  loss: 0.1956 (0.2082)  loss_classifier: 0.0541 (0.0594)  loss_box_reg: 0.1346 (0.1406)  loss_objectness: 0.0015 (0.0019)  loss_rpn_box_reg: 0.0047 (0.0064)  time: 0.9442  data: 0.0200  max mem: 4355
Epoch: [16]  [ 80/104]  eta: 0:00:23  lr: 0.000000  loss: 0.2074 (0.2137)  loss_classifier: 0.0587 (0.0613)  loss_box_reg: 0.1411 (0.1437)  loss_objectness: 0.0020 (0.0021)  loss_rpn_box_reg: 0.0062 (0.0066)  time: 0.9428  data: 0.0193  max mem: 4355
Epoch: [16]  [ 90/104]  eta: 0:00:13  lr: 0.000000  loss: 0.2074 (0.2114)  loss_classifier: 0.0661 (0.0608)  loss_box_reg: 0.1402 (0.1422)  loss_objectness: 0.0014 (0.0020)  loss_rpn_box_reg: 0.0049 (0.0063)  time: 0.9482  data: 0.0199  max mem: 4355
Epoch: [16]  [100/104]  eta: 0:00:03  lr: 0.000000  loss: 0.1901 (0.2125)  loss_classifier: 0.0544 (0.0607)  loss_box_reg: 0.1276 (0.1434)  loss_objectness: 0.0012 (0.0019)  loss_rpn_box_reg: 0.0048 (0.0064)  time: 0.9532  data: 0.0198  max mem: 4355
Epoch: [16]  [103/104]  eta: 0:00:00  lr: 0.000000  loss: 0.1901 (0.2129)  loss_classifier: 0.0544 (0.0608)  loss_box_reg: 0.1276 (0.1438)  loss_objectness: 0.0013 (0.0020)  loss_rpn_box_reg: 0.0052 (0.0064)  time: 0.9481  data: 0.0183  max mem: 4355
Epoch: [16] Total time: 0:01:40 (0.9645 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:29  model_time: 0.5039 (0.5039)  evaluator_time: 0.0272 (0.0272)  time: 1.1478  data: 0.5974  max mem: 4355
Test:  [25/26]  eta: 0:00:00  model_time: 0.4036 (0.4075)  evaluator_time: 0.0105 (0.0197)  time: 0.4528  data: 0.0203  max mem: 4355
Test: Total time: 0:00:12 (0.4817 s / it)
Averaged stats: model_time: 0.4036 (0.4075)  evaluator_time: 0.0105 (0.0197)
Accumulating evaluation results...
DONE (t=0.07s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.538
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.891
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.599
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.370
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.540
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.484
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.220
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.530
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.605
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.431
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.602
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.577
Epoch: [17]  [  0/104]  eta: 0:02:53  lr: 0.000000  loss: 0.2228 (0.2228)  loss_classifier: 0.0683 (0.0683)  loss_box_reg: 0.1485 (0.1485)  loss_objectness: 0.0005 (0.0005)  loss_rpn_box_reg: 0.0055 (0.0055)  time: 1.6703  data: 0.6723  max mem: 4355
Epoch: [17]  [ 10/104]  eta: 0:01:37  lr: 0.000000  loss: 0.2259 (0.2446)  loss_classifier: 0.0682 (0.0702)  loss_box_reg: 0.1485 (0.1652)  loss_objectness: 0.0024 (0.0025)  loss_rpn_box_reg: 0.0051 (0.0067)  time: 1.0412  data: 0.0821  max mem: 4355
Epoch: [17]  [ 20/104]  eta: 0:01:24  lr: 0.000000  loss: 0.2088 (0.2263)  loss_classifier: 0.0600 (0.0636)  loss_box_reg: 0.1369 (0.1543)  loss_objectness: 0.0013 (0.0021)  loss_rpn_box_reg: 0.0051 (0.0063)  time: 0.9775  data: 0.0209  max mem: 4355
Epoch: [17]  [ 30/104]  eta: 0:01:13  lr: 0.000000  loss: 0.1915 (0.2150)  loss_classifier: 0.0597 (0.0615)  loss_box_reg: 0.1356 (0.1455)  loss_objectness: 0.0010 (0.0021)  loss_rpn_box_reg: 0.0037 (0.0060)  time: 0.9745  data: 0.0200  max mem: 4355
Epoch: [17]  [ 40/104]  eta: 0:01:03  lr: 0.000000  loss: 0.1814 (0.2125)  loss_classifier: 0.0545 (0.0611)  loss_box_reg: 0.1115 (0.1434)  loss_objectness: 0.0017 (0.0021)  loss_rpn_box_reg: 0.0043 (0.0059)  time: 0.9677  data: 0.0219  max mem: 4355
Epoch: [17]  [ 50/104]  eta: 0:00:53  lr: 0.000000  loss: 0.1886 (0.2064)  loss_classifier: 0.0545 (0.0596)  loss_box_reg: 0.1244 (0.1392)  loss_objectness: 0.0015 (0.0020)  loss_rpn_box_reg: 0.0045 (0.0057)  time: 0.9561  data: 0.0217  max mem: 4355
Epoch: [17]  [ 60/104]  eta: 0:00:42  lr: 0.000000  loss: 0.1886 (0.2076)  loss_classifier: 0.0568 (0.0593)  loss_box_reg: 0.1260 (0.1406)  loss_objectness: 0.0010 (0.0019)  loss_rpn_box_reg: 0.0051 (0.0057)  time: 0.9437  data: 0.0201  max mem: 4355
Epoch: [17]  [ 70/104]  eta: 0:00:33  lr: 0.000000  loss: 0.1979 (0.2084)  loss_classifier: 0.0573 (0.0592)  loss_box_reg: 0.1360 (0.1410)  loss_objectness: 0.0009 (0.0019)  loss_rpn_box_reg: 0.0052 (0.0062)  time: 0.9444  data: 0.0211  max mem: 4355
Epoch: [17]  [ 80/104]  eta: 0:00:23  lr: 0.000000  loss: 0.2045 (0.2104)  loss_classifier: 0.0634 (0.0599)  loss_box_reg: 0.1410 (0.1421)  loss_objectness: 0.0016 (0.0021)  loss_rpn_box_reg: 0.0052 (0.0063)  time: 0.9502  data: 0.0218  max mem: 4355
Epoch: [17]  [ 90/104]  eta: 0:00:13  lr: 0.000000  loss: 0.2384 (0.2129)  loss_classifier: 0.0647 (0.0601)  loss_box_reg: 0.1636 (0.1442)  loss_objectness: 0.0020 (0.0021)  loss_rpn_box_reg: 0.0074 (0.0065)  time: 0.9483  data: 0.0199  max mem: 4355
Epoch: [17]  [100/104]  eta: 0:00:03  lr: 0.000000  loss: 0.2158 (0.2130)  loss_classifier: 0.0586 (0.0607)  loss_box_reg: 0.1437 (0.1438)  loss_objectness: 0.0016 (0.0021)  loss_rpn_box_reg: 0.0068 (0.0064)  time: 0.9540  data: 0.0199  max mem: 4355
Epoch: [17]  [103/104]  eta: 0:00:00  lr: 0.000000  loss: 0.2158 (0.2133)  loss_classifier: 0.0586 (0.0610)  loss_box_reg: 0.1389 (0.1439)  loss_objectness: 0.0013 (0.0020)  loss_rpn_box_reg: 0.0068 (0.0064)  time: 0.9535  data: 0.0194  max mem: 4355
Epoch: [17] Total time: 0:01:40 (0.9661 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:31  model_time: 0.4864 (0.4864)  evaluator_time: 0.0164 (0.0164)  time: 1.2006  data: 0.6887  max mem: 4355
Test:  [25/26]  eta: 0:00:00  model_time: 0.4035 (0.4075)  evaluator_time: 0.0103 (0.0186)  time: 0.4453  data: 0.0190  max mem: 4355
Test: Total time: 0:00:12 (0.4825 s / it)
Averaged stats: model_time: 0.4035 (0.4075)  evaluator_time: 0.0103 (0.0186)
Accumulating evaluation results...
DONE (t=0.08s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.538
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.891
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.599
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.370
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.540
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.484
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.220
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.530
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.605
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.431
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.602
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.577
Epoch: [18]  [  0/104]  eta: 0:02:44  lr: 0.000000  loss: 0.1811 (0.1811)  loss_classifier: 0.0471 (0.0471)  loss_box_reg: 0.1268 (0.1268)  loss_objectness: 0.0014 (0.0014)  loss_rpn_box_reg: 0.0057 (0.0057)  time: 1.5804  data: 0.5799  max mem: 4355
Epoch: [18]  [ 10/104]  eta: 0:01:35  lr: 0.000000  loss: 0.1901 (0.2078)  loss_classifier: 0.0491 (0.0559)  loss_box_reg: 0.1355 (0.1444)  loss_objectness: 0.0014 (0.0022)  loss_rpn_box_reg: 0.0054 (0.0053)  time: 1.0210  data: 0.0702  max mem: 4355
Epoch: [18]  [ 20/104]  eta: 0:01:24  lr: 0.000000  loss: 0.1901 (0.1995)  loss_classifier: 0.0491 (0.0550)  loss_box_reg: 0.1301 (0.1375)  loss_objectness: 0.0010 (0.0016)  loss_rpn_box_reg: 0.0051 (0.0054)  time: 0.9762  data: 0.0203  max mem: 4355
Epoch: [18]  [ 30/104]  eta: 0:01:13  lr: 0.000000  loss: 0.2086 (0.2101)  loss_classifier: 0.0564 (0.0594)  loss_box_reg: 0.1423 (0.1421)  loss_objectness: 0.0011 (0.0018)  loss_rpn_box_reg: 0.0058 (0.0067)  time: 0.9767  data: 0.0200  max mem: 4355
Epoch: [18]  [ 40/104]  eta: 0:01:03  lr: 0.000000  loss: 0.2099 (0.2119)  loss_classifier: 0.0646 (0.0605)  loss_box_reg: 0.1433 (0.1434)  loss_objectness: 0.0010 (0.0016)  loss_rpn_box_reg: 0.0058 (0.0065)  time: 0.9644  data: 0.0201  max mem: 4355
Epoch: [18]  [ 50/104]  eta: 0:00:52  lr: 0.000000  loss: 0.2033 (0.2132)  loss_classifier: 0.0574 (0.0613)  loss_box_reg: 0.1390 (0.1438)  loss_objectness: 0.0009 (0.0018)  loss_rpn_box_reg: 0.0048 (0.0063)  time: 0.9598  data: 0.0212  max mem: 4355
Epoch: [18]  [ 60/104]  eta: 0:00:42  lr: 0.000000  loss: 0.2164 (0.2134)  loss_classifier: 0.0625 (0.0611)  loss_box_reg: 0.1473 (0.1440)  loss_objectness: 0.0015 (0.0019)  loss_rpn_box_reg: 0.0052 (0.0064)  time: 0.9509  data: 0.0198  max mem: 4355
Epoch: [18]  [ 70/104]  eta: 0:00:32  lr: 0.000000  loss: 0.2271 (0.2201)  loss_classifier: 0.0684 (0.0633)  loss_box_reg: 0.1520 (0.1478)  loss_objectness: 0.0013 (0.0020)  loss_rpn_box_reg: 0.0072 (0.0070)  time: 0.9440  data: 0.0197  max mem: 4355
Epoch: [18]  [ 80/104]  eta: 0:00:23  lr: 0.000000  loss: 0.2077 (0.2164)  loss_classifier: 0.0689 (0.0623)  loss_box_reg: 0.1292 (0.1453)  loss_objectness: 0.0010 (0.0020)  loss_rpn_box_reg: 0.0062 (0.0067)  time: 0.9465  data: 0.0217  max mem: 4355
Epoch: [18]  [ 90/104]  eta: 0:00:13  lr: 0.000000  loss: 0.1816 (0.2148)  loss_classifier: 0.0550 (0.0620)  loss_box_reg: 0.1271 (0.1443)  loss_objectness: 0.0011 (0.0020)  loss_rpn_box_reg: 0.0045 (0.0066)  time: 0.9505  data: 0.0209  max mem: 4355
Epoch: [18]  [100/104]  eta: 0:00:03  lr: 0.000000  loss: 0.1953 (0.2134)  loss_classifier: 0.0550 (0.0614)  loss_box_reg: 0.1409 (0.1436)  loss_objectness: 0.0017 (0.0020)  loss_rpn_box_reg: 0.0050 (0.0064)  time: 0.9477  data: 0.0182  max mem: 4355
Epoch: [18]  [103/104]  eta: 0:00:00  lr: 0.000000  loss: 0.1915 (0.2134)  loss_classifier: 0.0503 (0.0613)  loss_box_reg: 0.1400 (0.1437)  loss_objectness: 0.0017 (0.0019)  loss_rpn_box_reg: 0.0050 (0.0064)  time: 0.9502  data: 0.0179  max mem: 4355
Epoch: [18] Total time: 0:01:40 (0.9646 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:44  model_time: 0.5206 (0.5206)  evaluator_time: 0.0464 (0.0464)  time: 1.6959  data: 1.0941  max mem: 4355
Test:  [25/26]  eta: 0:00:00  model_time: 0.4023 (0.4074)  evaluator_time: 0.0100 (0.0169)  time: 0.4413  data: 0.0183  max mem: 4355
Test: Total time: 0:00:12 (0.4958 s / it)
Averaged stats: model_time: 0.4023 (0.4074)  evaluator_time: 0.0100 (0.0169)
Accumulating evaluation results...
DONE (t=0.14s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.538
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.891
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.599
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.370
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.540
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.484
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.220
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.530
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.605
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.431
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.602
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.577
Epoch: [19]  [  0/104]  eta: 0:03:53  lr: 0.000000  loss: 0.1829 (0.1829)  loss_classifier: 0.0515 (0.0515)  loss_box_reg: 0.1262 (0.1262)  loss_objectness: 0.0006 (0.0006)  loss_rpn_box_reg: 0.0045 (0.0045)  time: 2.2405  data: 1.0624  max mem: 4355
Epoch: [19]  [ 10/104]  eta: 0:01:41  lr: 0.000000  loss: 0.2071 (0.2052)  loss_classifier: 0.0543 (0.0603)  loss_box_reg: 0.1446 (0.1373)  loss_objectness: 0.0013 (0.0020)  loss_rpn_box_reg: 0.0041 (0.0056)  time: 1.0810  data: 0.1122  max mem: 4355
Epoch: [19]  [ 20/104]  eta: 0:01:27  lr: 0.000000  loss: 0.2236 (0.2161)  loss_classifier: 0.0623 (0.0630)  loss_box_reg: 0.1482 (0.1448)  loss_objectness: 0.0013 (0.0020)  loss_rpn_box_reg: 0.0041 (0.0064)  time: 0.9863  data: 0.0227  max mem: 4355
Epoch: [19]  [ 30/104]  eta: 0:01:15  lr: 0.000000  loss: 0.2097 (0.2103)  loss_classifier: 0.0624 (0.0601)  loss_box_reg: 0.1477 (0.1420)  loss_objectness: 0.0012 (0.0021)  loss_rpn_box_reg: 0.0043 (0.0061)  time: 0.9885  data: 0.0240  max mem: 4355
Epoch: [19]  [ 40/104]  eta: 0:01:04  lr: 0.000000  loss: 0.1973 (0.2107)  loss_classifier: 0.0545 (0.0597)  loss_box_reg: 0.1385 (0.1430)  loss_objectness: 0.0010 (0.0020)  loss_rpn_box_reg: 0.0040 (0.0060)  time: 0.9628  data: 0.0196  max mem: 4355
Epoch: [19]  [ 50/104]  eta: 0:00:53  lr: 0.000000  loss: 0.2070 (0.2117)  loss_classifier: 0.0605 (0.0604)  loss_box_reg: 0.1422 (0.1434)  loss_objectness: 0.0011 (0.0020)  loss_rpn_box_reg: 0.0048 (0.0059)  time: 0.9559  data: 0.0208  max mem: 4355
Epoch: [19]  [ 60/104]  eta: 0:00:43  lr: 0.000000  loss: 0.2082 (0.2123)  loss_classifier: 0.0610 (0.0605)  loss_box_reg: 0.1472 (0.1438)  loss_objectness: 0.0013 (0.0020)  loss_rpn_box_reg: 0.0052 (0.0059)  time: 0.9552  data: 0.0220  max mem: 4355
Epoch: [19]  [ 70/104]  eta: 0:00:33  lr: 0.000000  loss: 0.2055 (0.2094)  loss_classifier: 0.0566 (0.0601)  loss_box_reg: 0.1378 (0.1415)  loss_objectness: 0.0013 (0.0020)  loss_rpn_box_reg: 0.0051 (0.0058)  time: 0.9512  data: 0.0225  max mem: 4355
Epoch: [19]  [ 80/104]  eta: 0:00:23  lr: 0.000000  loss: 0.1907 (0.2100)  loss_classifier: 0.0542 (0.0603)  loss_box_reg: 0.1284 (0.1417)  loss_objectness: 0.0015 (0.0020)  loss_rpn_box_reg: 0.0061 (0.0060)  time: 0.9499  data: 0.0225  max mem: 4355
Epoch: [19]  [ 90/104]  eta: 0:00:13  lr: 0.000000  loss: 0.2205 (0.2123)  loss_classifier: 0.0542 (0.0612)  loss_box_reg: 0.1497 (0.1430)  loss_objectness: 0.0016 (0.0020)  loss_rpn_box_reg: 0.0061 (0.0061)  time: 0.9531  data: 0.0213  max mem: 4355
Epoch: [19]  [100/104]  eta: 0:00:03  lr: 0.000000  loss: 0.2263 (0.2144)  loss_classifier: 0.0658 (0.0623)  loss_box_reg: 0.1501 (0.1435)  loss_objectness: 0.0020 (0.0021)  loss_rpn_box_reg: 0.0061 (0.0065)  time: 0.9510  data: 0.0190  max mem: 4355
Epoch: [19]  [103/104]  eta: 0:00:00  lr: 0.000000  loss: 0.2239 (0.2140)  loss_classifier: 0.0591 (0.0621)  loss_box_reg: 0.1518 (0.1434)  loss_objectness: 0.0020 (0.0021)  loss_rpn_box_reg: 0.0061 (0.0064)  time: 0.9546  data: 0.0190  max mem: 4355
Epoch: [19] Total time: 0:01:41 (0.9745 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:38  model_time: 0.4969 (0.4969)  evaluator_time: 0.0347 (0.0347)  time: 1.4771  data: 0.9138  max mem: 4355
Test:  [25/26]  eta: 0:00:00  model_time: 0.4027 (0.4067)  evaluator_time: 0.0105 (0.0176)  time: 0.4442  data: 0.0181  max mem: 4355
Test: Total time: 0:00:12 (0.4899 s / it)
Averaged stats: model_time: 0.4027 (0.4067)  evaluator_time: 0.0105 (0.0176)
Accumulating evaluation results...
DONE (t=0.14s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.538
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.891
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.599
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.370
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.540
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.484
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.220
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.530
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.605
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.431
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.602
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.577

#save adamn
import pickle
Filename = "FRCNNsgd.pkl"
# Define the file path where you want to save the model
filename = "/content/drive/MyDrive/dataset/FRCNNsgd.pkl"

# Save the model to the specified file path
torch.save(model.state_dict(), filename)
# Save the Modle to file in the current working directory
with open(Filename, 'wb') as file:
    pickle.dump(model, file)
# Load the Model back from file
with open(Filename, 'rb') as file:
    model = pickle.load(file)
model

FasterRCNN(
  (transform): GeneralizedRCNNTransform(
      Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
      Resize(min_size=(800,), max_size=1333, mode='bilinear')
  )
  (backbone): BackboneWithFPN(
    (body): IntermediateLayerGetter(
      (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)
      (bn1): FrozenBatchNorm2d(64, eps=0.0)
      (relu): ReLU(inplace=True)
      (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)
      (layer1): Sequential(
        (0): Bottleneck(
          (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(64, eps=0.0)
          (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(64, eps=0.0)
          (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(256, eps=0.0)
          (relu): ReLU(inplace=True)
          (downsample): Sequential(
            (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
            (1): FrozenBatchNorm2d(256, eps=0.0)
          )
        )
        (1): Bottleneck(
          (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(64, eps=0.0)
          (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(64, eps=0.0)
          (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(256, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (2): Bottleneck(
          (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(64, eps=0.0)
          (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(64, eps=0.0)
          (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(256, eps=0.0)
          (relu): ReLU(inplace=True)
        )
      )
      (layer2): Sequential(
        (0): Bottleneck(
          (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(128, eps=0.0)
          (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(128, eps=0.0)
          (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(512, eps=0.0)
          (relu): ReLU(inplace=True)
          (downsample): Sequential(
            (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)
            (1): FrozenBatchNorm2d(512, eps=0.0)
          )
        )
        (1): Bottleneck(
          (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(128, eps=0.0)
          (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(128, eps=0.0)
          (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(512, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (2): Bottleneck(
          (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(128, eps=0.0)
          (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(128, eps=0.0)
          (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(512, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (3): Bottleneck(
          (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(128, eps=0.0)
          (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(128, eps=0.0)
          (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(512, eps=0.0)
          (relu): ReLU(inplace=True)
        )
      )
      (layer3): Sequential(
        (0): Bottleneck(
          (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
          (downsample): Sequential(
            (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False)
            (1): FrozenBatchNorm2d(1024, eps=0.0)
          )
        )
        (1): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (2): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (3): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (4): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (5): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
        )
      )
      (layer4): Sequential(
        (0): Bottleneck(
          (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(512, eps=0.0)
          (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(512, eps=0.0)
          (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(2048, eps=0.0)
          (relu): ReLU(inplace=True)
          (downsample): Sequential(
            (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False)
            (1): FrozenBatchNorm2d(2048, eps=0.0)
          )
        )
        (1): Bottleneck(
          (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(512, eps=0.0)
          (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(512, eps=0.0)
          (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(2048, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (2): Bottleneck(
          (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(512, eps=0.0)
          (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(512, eps=0.0)
          (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(2048, eps=0.0)
          (relu): ReLU(inplace=True)
        )
      )
    )
    (fpn): FeaturePyramidNetwork(
      (inner_blocks): ModuleList(
        (0): Conv2dNormActivation(
          (0): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1))
        )
        (1): Conv2dNormActivation(
          (0): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1))
        )
        (2): Conv2dNormActivation(
          (0): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1))
        )
        (3): Conv2dNormActivation(
          (0): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1))
        )
      )
      (layer_blocks): ModuleList(
        (0-3): 4 x Conv2dNormActivation(
          (0): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
        )
      )
      (extra_blocks): LastLevelMaxPool()
    )
  )
  (rpn): RegionProposalNetwork(
    (anchor_generator): AnchorGenerator()
    (head): RPNHead(
      (conv): Sequential(
        (0): Conv2dNormActivation(
          (0): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
          (1): ReLU(inplace=True)
        )
      )
      (cls_logits): Conv2d(256, 3, kernel_size=(1, 1), stride=(1, 1))
      (bbox_pred): Conv2d(256, 12, kernel_size=(1, 1), stride=(1, 1))
    )
  )
  (roi_heads): RoIHeads(
    (box_roi_pool): MultiScaleRoIAlign(featmap_names=['0', '1', '2', '3'], output_size=(7, 7), sampling_ratio=2)
    (box_head): TwoMLPHead(
      (fc6): Linear(in_features=12544, out_features=1024, bias=True)
      (fc7): Linear(in_features=1024, out_features=1024, bias=True)
    )
    (box_predictor): FastRCNNPredictor(
      (cls_score): Linear(in_features=1024, out_features=11, bias=True)
      (bbox_pred): Linear(in_features=1024, out_features=44, bias=True)
    )
  )
)

# to train on GPU if selected
device = torch.device('cuda') if torch.cuda.is_available() else torch.device('cpu')

# number of classes
num_classes = 11

# get the model using our helper function
model = get_object_detection_model(num_classes)

# move model to the right device
model.to(device)

# construct an optimizer
params = [p for p in model.parameters() if p.requires_grad]
optimizer = torch.optim.Adam(params, lr=0.001, weight_decay=0.0005)

# and a learning rate scheduler which decreases the learning rate by
# 10x every 3 epochs
lr_scheduler = torch.optim.lr_scheduler.StepLR(optimizer,
                                               step_size=3,
                                               gamma=0.1)

# training for 8 epochs # adam
num_epochs = 15

for epoch in range(num_epochs):
    # training for one epoch
    train_one_epoch(model, optimizer, data_loader, device, epoch, print_freq=10)
    # update the learning rate
    lr_scheduler.step()
    # evaluate on the test dataset
    evaluate(model, data_loader_test, device=device)

Epoch: [0]  [  0/104]  eta: 0:03:27  lr: 0.000011  loss: 3.4412 (3.4412)  loss_classifier: 2.6871 (2.6871)  loss_box_reg: 0.2420 (0.2420)  loss_objectness: 0.4842 (0.4842)  loss_rpn_box_reg: 0.0278 (0.0278)  time: 1.9959  data: 0.8924  max mem: 6508
Epoch: [0]  [ 10/104]  eta: 0:01:51  lr: 0.000108  loss: 1.9666 (2.0528)  loss_classifier: 1.1984 (1.5021)  loss_box_reg: 0.3116 (0.3519)  loss_objectness: 0.1556 (0.1792)  loss_rpn_box_reg: 0.0162 (0.0196)  time: 1.1900  data: 0.0974  max mem: 6508
Epoch: [0]  [ 20/104]  eta: 0:01:37  lr: 0.000205  loss: 1.0744 (1.5374)  loss_classifier: 0.5573 (1.0176)  loss_box_reg: 0.3418 (0.3596)  loss_objectness: 0.1066 (0.1406)  loss_rpn_box_reg: 0.0162 (0.0196)  time: 1.1201  data: 0.0198  max mem: 6508
Epoch: [0]  [ 30/104]  eta: 0:01:23  lr: 0.000302  loss: 0.8417 (1.3277)  loss_classifier: 0.4224 (0.8269)  loss_box_reg: 0.3510 (0.3689)  loss_objectness: 0.0588 (0.1136)  loss_rpn_box_reg: 0.0132 (0.0184)  time: 1.1032  data: 0.0203  max mem: 6508
Epoch: [0]  [ 40/104]  eta: 0:01:11  lr: 0.000399  loss: 0.8124 (1.2217)  loss_classifier: 0.4127 (0.7357)  loss_box_reg: 0.3445 (0.3706)  loss_objectness: 0.0425 (0.0973)  loss_rpn_box_reg: 0.0132 (0.0182)  time: 1.0606  data: 0.0197  max mem: 6508
Epoch: [0]  [ 50/104]  eta: 0:00:59  lr: 0.000496  loss: 0.8505 (1.1649)  loss_classifier: 0.4300 (0.6826)  loss_box_reg: 0.3445 (0.3667)  loss_objectness: 0.0415 (0.0961)  loss_rpn_box_reg: 0.0172 (0.0196)  time: 1.0428  data: 0.0225  max mem: 6508
Epoch: [0]  [ 60/104]  eta: 0:00:47  lr: 0.000593  loss: 0.9567 (1.1331)  loss_classifier: 0.4300 (0.6512)  loss_box_reg: 0.3505 (0.3687)  loss_objectness: 0.0581 (0.0926)  loss_rpn_box_reg: 0.0223 (0.0206)  time: 1.0324  data: 0.0233  max mem: 6508
Epoch: [0]  [ 70/104]  eta: 0:00:36  lr: 0.000690  loss: 0.8243 (1.0775)  loss_classifier: 0.3951 (0.6126)  loss_box_reg: 0.2874 (0.3552)  loss_objectness: 0.0613 (0.0892)  loss_rpn_box_reg: 0.0214 (0.0204)  time: 1.0229  data: 0.0206  max mem: 6508
Epoch: [0]  [ 80/104]  eta: 0:00:25  lr: 0.000787  loss: 0.8379 (1.0610)  loss_classifier: 0.3951 (0.5918)  loss_box_reg: 0.3641 (0.3622)  loss_objectness: 0.0537 (0.0854)  loss_rpn_box_reg: 0.0173 (0.0216)  time: 1.0280  data: 0.0188  max mem: 6508
Epoch: [0]  [ 90/104]  eta: 0:00:14  lr: 0.000884  loss: 0.8269 (1.0324)  loss_classifier: 0.3935 (0.5685)  loss_box_reg: 0.3835 (0.3585)  loss_objectness: 0.0555 (0.0838)  loss_rpn_box_reg: 0.0216 (0.0215)  time: 1.0435  data: 0.0205  max mem: 6508
Epoch: [0]  [100/104]  eta: 0:00:04  lr: 0.000981  loss: 0.7398 (1.0072)  loss_classifier: 0.3464 (0.5445)  loss_box_reg: 0.3031 (0.3547)  loss_objectness: 0.0724 (0.0851)  loss_rpn_box_reg: 0.0199 (0.0228)  time: 1.0630  data: 0.0221  max mem: 6508
Epoch: [0]  [103/104]  eta: 0:00:01  lr: 0.001000  loss: 0.7319 (0.9981)  loss_classifier: 0.3661 (0.5396)  loss_box_reg: 0.2775 (0.3502)  loss_objectness: 0.0815 (0.0854)  loss_rpn_box_reg: 0.0199 (0.0229)  time: 1.0630  data: 0.0218  max mem: 6508
Epoch: [0] Total time: 0:01:51 (1.0705 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:32  model_time: 0.5127 (0.5127)  evaluator_time: 0.0156 (0.0156)  time: 1.2586  data: 0.7145  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4363 (0.4386)  evaluator_time: 0.0207 (0.0235)  time: 0.4886  data: 0.0211  max mem: 6508
Test: Total time: 0:00:13 (0.5200 s / it)
Averaged stats: model_time: 0.4363 (0.4386)  evaluator_time: 0.0207 (0.0235)
Accumulating evaluation results...
DONE (t=0.16s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.086
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.227
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.041
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.082
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.079
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.079
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.060
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.138
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.180
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.206
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.199
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.088
Epoch: [1]  [  0/104]  eta: 0:03:21  lr: 0.001000  loss: 0.8187 (0.8187)  loss_classifier: 0.3262 (0.3262)  loss_box_reg: 0.3955 (0.3955)  loss_objectness: 0.0776 (0.0776)  loss_rpn_box_reg: 0.0195 (0.0195)  time: 1.9329  data: 0.8546  max mem: 6508
Epoch: [1]  [ 10/104]  eta: 0:01:47  lr: 0.001000  loss: 0.6260 (0.6550)  loss_classifier: 0.2835 (0.2900)  loss_box_reg: 0.2487 (0.2620)  loss_objectness: 0.0706 (0.0771)  loss_rpn_box_reg: 0.0216 (0.0259)  time: 1.1428  data: 0.0923  max mem: 6508
Epoch: [1]  [ 20/104]  eta: 0:01:33  lr: 0.001000  loss: 0.6369 (0.6688)  loss_classifier: 0.2835 (0.2994)  loss_box_reg: 0.2845 (0.2835)  loss_objectness: 0.0516 (0.0622)  loss_rpn_box_reg: 0.0199 (0.0237)  time: 1.0669  data: 0.0174  max mem: 6508
Epoch: [1]  [ 30/104]  eta: 0:01:20  lr: 0.001000  loss: 0.6915 (0.6925)  loss_classifier: 0.2926 (0.3079)  loss_box_reg: 0.3091 (0.3023)  loss_objectness: 0.0429 (0.0586)  loss_rpn_box_reg: 0.0173 (0.0237)  time: 1.0603  data: 0.0194  max mem: 6508
Epoch: [1]  [ 40/104]  eta: 0:01:09  lr: 0.001000  loss: 0.7575 (0.7170)  loss_classifier: 0.2951 (0.3153)  loss_box_reg: 0.3427 (0.3221)  loss_objectness: 0.0464 (0.0553)  loss_rpn_box_reg: 0.0196 (0.0243)  time: 1.0481  data: 0.0203  max mem: 6508
Epoch: [1]  [ 50/104]  eta: 0:00:57  lr: 0.001000  loss: 0.7855 (0.7342)  loss_classifier: 0.3421 (0.3251)  loss_box_reg: 0.3934 (0.3338)  loss_objectness: 0.0324 (0.0501)  loss_rpn_box_reg: 0.0207 (0.0252)  time: 1.0417  data: 0.0209  max mem: 6508
Epoch: [1]  [ 60/104]  eta: 0:00:46  lr: 0.001000  loss: 0.7067 (0.7303)  loss_classifier: 0.3367 (0.3236)  loss_box_reg: 0.3516 (0.3330)  loss_objectness: 0.0316 (0.0498)  loss_rpn_box_reg: 0.0179 (0.0239)  time: 1.0339  data: 0.0210  max mem: 6508
Epoch: [1]  [ 70/104]  eta: 0:00:36  lr: 0.001000  loss: 0.6577 (0.7207)  loss_classifier: 0.2653 (0.3187)  loss_box_reg: 0.2943 (0.3288)  loss_objectness: 0.0441 (0.0500)  loss_rpn_box_reg: 0.0164 (0.0232)  time: 1.0331  data: 0.0216  max mem: 6508
Epoch: [1]  [ 80/104]  eta: 0:00:25  lr: 0.001000  loss: 0.6020 (0.7042)  loss_classifier: 0.2453 (0.3072)  loss_box_reg: 0.2700 (0.3235)  loss_objectness: 0.0439 (0.0500)  loss_rpn_box_reg: 0.0216 (0.0235)  time: 1.0369  data: 0.0223  max mem: 6508
Epoch: [1]  [ 90/104]  eta: 0:00:14  lr: 0.001000  loss: 0.5815 (0.6954)  loss_classifier: 0.2143 (0.3037)  loss_box_reg: 0.2780 (0.3215)  loss_objectness: 0.0288 (0.0475)  loss_rpn_box_reg: 0.0215 (0.0227)  time: 1.0432  data: 0.0225  max mem: 6508
Epoch: [1]  [100/104]  eta: 0:00:04  lr: 0.001000  loss: 0.6848 (0.6985)  loss_classifier: 0.2656 (0.3040)  loss_box_reg: 0.3175 (0.3224)  loss_objectness: 0.0325 (0.0491)  loss_rpn_box_reg: 0.0176 (0.0231)  time: 1.0425  data: 0.0211  max mem: 6508
Epoch: [1]  [103/104]  eta: 0:00:01  lr: 0.001000  loss: 0.5815 (0.6889)  loss_classifier: 0.2348 (0.2999)  loss_box_reg: 0.2664 (0.3178)  loss_objectness: 0.0311 (0.0485)  loss_rpn_box_reg: 0.0149 (0.0227)  time: 1.0401  data: 0.0202  max mem: 6508
Epoch: [1] Total time: 0:01:49 (1.0551 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:34  model_time: 0.5233 (0.5233)  evaluator_time: 0.0307 (0.0307)  time: 1.3337  data: 0.7621  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4390 (0.4415)  evaluator_time: 0.0232 (0.0258)  time: 0.4913  data: 0.0191  max mem: 6508
Test: Total time: 0:00:13 (0.5284 s / it)
Averaged stats: model_time: 0.4390 (0.4415)  evaluator_time: 0.0232 (0.0258)
Accumulating evaluation results...
DONE (t=0.25s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.209
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.479
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.124
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.186
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.257
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.087
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.108
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.303
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.351
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.316
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.396
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.183
Epoch: [2]  [  0/104]  eta: 0:03:09  lr: 0.001000  loss: 0.2370 (0.2370)  loss_classifier: 0.0976 (0.0976)  loss_box_reg: 0.1113 (0.1113)  loss_objectness: 0.0123 (0.0123)  loss_rpn_box_reg: 0.0158 (0.0158)  time: 1.8269  data: 0.7389  max mem: 6508
Epoch: [2]  [ 10/104]  eta: 0:01:47  lr: 0.001000  loss: 0.5648 (0.5501)  loss_classifier: 0.1963 (0.2145)  loss_box_reg: 0.2968 (0.2883)  loss_objectness: 0.0258 (0.0299)  loss_rpn_box_reg: 0.0164 (0.0173)  time: 1.1384  data: 0.0856  max mem: 6508
Epoch: [2]  [ 20/104]  eta: 0:01:32  lr: 0.001000  loss: 0.5648 (0.5797)  loss_classifier: 0.2199 (0.2261)  loss_box_reg: 0.3092 (0.3060)  loss_objectness: 0.0264 (0.0316)  loss_rpn_box_reg: 0.0157 (0.0161)  time: 1.0709  data: 0.0207  max mem: 6508
Epoch: [2]  [ 30/104]  eta: 0:01:20  lr: 0.001000  loss: 0.5875 (0.5984)  loss_classifier: 0.2595 (0.2481)  loss_box_reg: 0.3092 (0.3002)  loss_objectness: 0.0277 (0.0320)  loss_rpn_box_reg: 0.0142 (0.0181)  time: 1.0615  data: 0.0213  max mem: 6508
Epoch: [2]  [ 40/104]  eta: 0:01:08  lr: 0.001000  loss: 0.6245 (0.6103)  loss_classifier: 0.3000 (0.2579)  loss_box_reg: 0.3041 (0.3047)  loss_objectness: 0.0231 (0.0300)  loss_rpn_box_reg: 0.0142 (0.0176)  time: 1.0455  data: 0.0207  max mem: 6508
Epoch: [2]  [ 50/104]  eta: 0:00:57  lr: 0.001000  loss: 0.6245 (0.6336)  loss_classifier: 0.2793 (0.2761)  loss_box_reg: 0.3083 (0.3066)  loss_objectness: 0.0225 (0.0332)  loss_rpn_box_reg: 0.0139 (0.0177)  time: 1.0367  data: 0.0215  max mem: 6508
Epoch: [2]  [ 60/104]  eta: 0:00:46  lr: 0.001000  loss: 0.6124 (0.6221)  loss_classifier: 0.2727 (0.2745)  loss_box_reg: 0.2801 (0.2973)  loss_objectness: 0.0352 (0.0331)  loss_rpn_box_reg: 0.0139 (0.0173)  time: 1.0268  data: 0.0214  max mem: 6508
Epoch: [2]  [ 70/104]  eta: 0:00:35  lr: 0.001000  loss: 0.6184 (0.6230)  loss_classifier: 0.2489 (0.2697)  loss_box_reg: 0.2801 (0.2991)  loss_objectness: 0.0333 (0.0354)  loss_rpn_box_reg: 0.0180 (0.0188)  time: 1.0177  data: 0.0197  max mem: 6508
Epoch: [2]  [ 80/104]  eta: 0:00:25  lr: 0.001000  loss: 0.6771 (0.6347)  loss_classifier: 0.2657 (0.2747)  loss_box_reg: 0.3387 (0.3036)  loss_objectness: 0.0374 (0.0368)  loss_rpn_box_reg: 0.0227 (0.0196)  time: 1.0247  data: 0.0205  max mem: 6508
Epoch: [2]  [ 90/104]  eta: 0:00:14  lr: 0.001000  loss: 0.6771 (0.6362)  loss_classifier: 0.2679 (0.2732)  loss_box_reg: 0.3083 (0.3045)  loss_objectness: 0.0435 (0.0381)  loss_rpn_box_reg: 0.0217 (0.0204)  time: 1.0385  data: 0.0222  max mem: 6508
Epoch: [2]  [100/104]  eta: 0:00:04  lr: 0.001000  loss: 0.6658 (0.6392)  loss_classifier: 0.2378 (0.2721)  loss_box_reg: 0.3310 (0.3084)  loss_objectness: 0.0319 (0.0376)  loss_rpn_box_reg: 0.0217 (0.0211)  time: 1.0424  data: 0.0209  max mem: 6508
Epoch: [2]  [103/104]  eta: 0:00:01  lr: 0.001000  loss: 0.6658 (0.6404)  loss_classifier: 0.2378 (0.2741)  loss_box_reg: 0.3307 (0.3077)  loss_objectness: 0.0302 (0.0376)  loss_rpn_box_reg: 0.0200 (0.0211)  time: 1.0437  data: 0.0205  max mem: 6508
Epoch: [2] Total time: 0:01:49 (1.0507 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:34  model_time: 0.5209 (0.5209)  evaluator_time: 0.0362 (0.0362)  time: 1.3311  data: 0.7572  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4389 (0.4390)  evaluator_time: 0.0220 (0.0398)  time: 0.5097  data: 0.0200  max mem: 6508
Test: Total time: 0:00:14 (0.5397 s / it)
Averaged stats: model_time: 0.4389 (0.4390)  evaluator_time: 0.0220 (0.0398)
Accumulating evaluation results...
DONE (t=0.16s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.212
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.534
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.122
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.130
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.247
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.132
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.111
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.311
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.356
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.289
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.416
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.246
Epoch: [3]  [  0/104]  eta: 0:03:08  lr: 0.000100  loss: 0.6804 (0.6804)  loss_classifier: 0.2065 (0.2065)  loss_box_reg: 0.3854 (0.3854)  loss_objectness: 0.0661 (0.0661)  loss_rpn_box_reg: 0.0224 (0.0224)  time: 1.8167  data: 0.6963  max mem: 6508
Epoch: [3]  [ 10/104]  eta: 0:01:47  lr: 0.000100  loss: 0.6202 (0.5907)  loss_classifier: 0.2539 (0.2473)  loss_box_reg: 0.2812 (0.2938)  loss_objectness: 0.0324 (0.0320)  loss_rpn_box_reg: 0.0170 (0.0176)  time: 1.1409  data: 0.0849  max mem: 6508
Epoch: [3]  [ 20/104]  eta: 0:01:33  lr: 0.000100  loss: 0.5724 (0.5738)  loss_classifier: 0.2272 (0.2427)  loss_box_reg: 0.2722 (0.2854)  loss_objectness: 0.0196 (0.0267)  loss_rpn_box_reg: 0.0166 (0.0190)  time: 1.0747  data: 0.0220  max mem: 6508
Epoch: [3]  [ 30/104]  eta: 0:01:20  lr: 0.000100  loss: 0.5440 (0.5548)  loss_classifier: 0.2248 (0.2326)  loss_box_reg: 0.2523 (0.2793)  loss_objectness: 0.0185 (0.0241)  loss_rpn_box_reg: 0.0143 (0.0189)  time: 1.0620  data: 0.0196  max mem: 6508
Epoch: [3]  [ 40/104]  eta: 0:01:08  lr: 0.000100  loss: 0.4058 (0.5266)  loss_classifier: 0.1643 (0.2222)  loss_box_reg: 0.2174 (0.2646)  loss_objectness: 0.0146 (0.0218)  loss_rpn_box_reg: 0.0138 (0.0180)  time: 1.0424  data: 0.0191  max mem: 6508
Epoch: [3]  [ 50/104]  eta: 0:00:57  lr: 0.000100  loss: 0.4355 (0.5124)  loss_classifier: 0.1803 (0.2144)  loss_box_reg: 0.2227 (0.2604)  loss_objectness: 0.0157 (0.0206)  loss_rpn_box_reg: 0.0144 (0.0171)  time: 1.0329  data: 0.0197  max mem: 6508
Epoch: [3]  [ 60/104]  eta: 0:00:46  lr: 0.000100  loss: 0.4576 (0.5058)  loss_classifier: 0.1871 (0.2110)  loss_box_reg: 0.2408 (0.2578)  loss_objectness: 0.0167 (0.0200)  loss_rpn_box_reg: 0.0136 (0.0169)  time: 1.0261  data: 0.0210  max mem: 6508
Epoch: [3]  [ 70/104]  eta: 0:00:35  lr: 0.000100  loss: 0.4142 (0.4890)  loss_classifier: 0.1623 (0.2011)  loss_box_reg: 0.2172 (0.2521)  loss_objectness: 0.0118 (0.0188)  loss_rpn_box_reg: 0.0135 (0.0169)  time: 1.0213  data: 0.0208  max mem: 6508
Epoch: [3]  [ 80/104]  eta: 0:00:25  lr: 0.000100  loss: 0.3602 (0.4781)  loss_classifier: 0.1426 (0.1944)  loss_box_reg: 0.2080 (0.2488)  loss_objectness: 0.0113 (0.0181)  loss_rpn_box_reg: 0.0128 (0.0168)  time: 1.0247  data: 0.0192  max mem: 6508
Epoch: [3]  [ 90/104]  eta: 0:00:14  lr: 0.000100  loss: 0.3589 (0.4657)  loss_classifier: 0.1411 (0.1882)  loss_box_reg: 0.2026 (0.2441)  loss_objectness: 0.0113 (0.0174)  loss_rpn_box_reg: 0.0097 (0.0161)  time: 1.0343  data: 0.0197  max mem: 6508
Epoch: [3]  [100/104]  eta: 0:00:04  lr: 0.000100  loss: 0.3456 (0.4548)  loss_classifier: 0.1264 (0.1822)  loss_box_reg: 0.2009 (0.2405)  loss_objectness: 0.0081 (0.0166)  loss_rpn_box_reg: 0.0088 (0.0155)  time: 1.0395  data: 0.0200  max mem: 6508
Epoch: [3]  [103/104]  eta: 0:00:01  lr: 0.000100  loss: 0.3425 (0.4524)  loss_classifier: 0.1245 (0.1807)  loss_box_reg: 0.1985 (0.2399)  loss_objectness: 0.0081 (0.0163)  loss_rpn_box_reg: 0.0088 (0.0154)  time: 1.0407  data: 0.0198  max mem: 6508
Epoch: [3] Total time: 0:01:49 (1.0501 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:34  model_time: 0.4908 (0.4908)  evaluator_time: 0.0537 (0.0537)  time: 1.3125  data: 0.7402  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4388 (0.4388)  evaluator_time: 0.0189 (0.0237)  time: 0.4892  data: 0.0209  max mem: 6508
Test: Total time: 0:00:13 (0.5227 s / it)
Averaged stats: model_time: 0.4388 (0.4388)  evaluator_time: 0.0189 (0.0237)
Accumulating evaluation results...
DONE (t=0.15s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.408
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.733
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.421
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.316
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.459
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.320
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.176
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.483
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.545
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.438
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.588
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.441
Epoch: [4]  [  0/104]  eta: 0:03:25  lr: 0.000100  loss: 0.5567 (0.5567)  loss_classifier: 0.2793 (0.2793)  loss_box_reg: 0.2358 (0.2358)  loss_objectness: 0.0143 (0.0143)  loss_rpn_box_reg: 0.0273 (0.0273)  time: 1.9788  data: 0.8958  max mem: 6508
Epoch: [4]  [ 10/104]  eta: 0:01:47  lr: 0.000100  loss: 0.3472 (0.3862)  loss_classifier: 0.1236 (0.1439)  loss_box_reg: 0.2126 (0.2210)  loss_objectness: 0.0074 (0.0085)  loss_rpn_box_reg: 0.0107 (0.0127)  time: 1.1458  data: 0.0970  max mem: 6508
Epoch: [4]  [ 20/104]  eta: 0:01:33  lr: 0.000100  loss: 0.3464 (0.3679)  loss_classifier: 0.1226 (0.1420)  loss_box_reg: 0.2008 (0.2060)  loss_objectness: 0.0074 (0.0083)  loss_rpn_box_reg: 0.0090 (0.0116)  time: 1.0733  data: 0.0188  max mem: 6508
Epoch: [4]  [ 30/104]  eta: 0:01:21  lr: 0.000100  loss: 0.3354 (0.3654)  loss_classifier: 0.1258 (0.1417)  loss_box_reg: 0.1787 (0.2014)  loss_objectness: 0.0095 (0.0106)  loss_rpn_box_reg: 0.0081 (0.0116)  time: 1.0694  data: 0.0209  max mem: 6508
Epoch: [4]  [ 40/104]  eta: 0:01:09  lr: 0.000100  loss: 0.3787 (0.3697)  loss_classifier: 0.1354 (0.1431)  loss_box_reg: 0.2111 (0.2041)  loss_objectness: 0.0086 (0.0102)  loss_rpn_box_reg: 0.0104 (0.0123)  time: 1.0458  data: 0.0206  max mem: 6508
Epoch: [4]  [ 50/104]  eta: 0:00:57  lr: 0.000100  loss: 0.3625 (0.3595)  loss_classifier: 0.1303 (0.1383)  loss_box_reg: 0.2021 (0.1997)  loss_objectness: 0.0069 (0.0097)  loss_rpn_box_reg: 0.0091 (0.0119)  time: 1.0313  data: 0.0203  max mem: 6508
Epoch: [4]  [ 60/104]  eta: 0:00:46  lr: 0.000100  loss: 0.3580 (0.3681)  loss_classifier: 0.1295 (0.1408)  loss_box_reg: 0.2104 (0.2048)  loss_objectness: 0.0088 (0.0104)  loss_rpn_box_reg: 0.0093 (0.0122)  time: 1.0259  data: 0.0217  max mem: 6508
Epoch: [4]  [ 70/104]  eta: 0:00:35  lr: 0.000100  loss: 0.4037 (0.3660)  loss_classifier: 0.1394 (0.1393)  loss_box_reg: 0.2228 (0.2046)  loss_objectness: 0.0093 (0.0102)  loss_rpn_box_reg: 0.0093 (0.0119)  time: 1.0230  data: 0.0207  max mem: 6508
Epoch: [4]  [ 80/104]  eta: 0:00:25  lr: 0.000100  loss: 0.3903 (0.3696)  loss_classifier: 0.1387 (0.1403)  loss_box_reg: 0.2209 (0.2077)  loss_objectness: 0.0059 (0.0098)  loss_rpn_box_reg: 0.0095 (0.0119)  time: 1.0291  data: 0.0199  max mem: 6508
Epoch: [4]  [ 90/104]  eta: 0:00:14  lr: 0.000100  loss: 0.3678 (0.3685)  loss_classifier: 0.1358 (0.1386)  loss_box_reg: 0.2143 (0.2082)  loss_objectness: 0.0059 (0.0097)  loss_rpn_box_reg: 0.0107 (0.0120)  time: 1.0391  data: 0.0204  max mem: 6508
Epoch: [4]  [100/104]  eta: 0:00:04  lr: 0.000100  loss: 0.3634 (0.3687)  loss_classifier: 0.1216 (0.1377)  loss_box_reg: 0.2001 (0.2093)  loss_objectness: 0.0059 (0.0098)  loss_rpn_box_reg: 0.0103 (0.0120)  time: 1.0395  data: 0.0189  max mem: 6508
Epoch: [4]  [103/104]  eta: 0:00:01  lr: 0.000100  loss: 0.3103 (0.3659)  loss_classifier: 0.1062 (0.1370)  loss_box_reg: 0.1927 (0.2073)  loss_objectness: 0.0061 (0.0099)  loss_rpn_box_reg: 0.0099 (0.0118)  time: 1.0396  data: 0.0185  max mem: 6508
Epoch: [4] Total time: 0:01:49 (1.0528 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:32  model_time: 0.5330 (0.5330)  evaluator_time: 0.0306 (0.0306)  time: 1.2520  data: 0.6645  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4393 (0.4412)  evaluator_time: 0.0190 (0.0213)  time: 0.4881  data: 0.0204  max mem: 6508
Test: Total time: 0:00:13 (0.5197 s / it)
Averaged stats: model_time: 0.4393 (0.4412)  evaluator_time: 0.0190 (0.0213)
Accumulating evaluation results...
DONE (t=0.13s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.429
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.779
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.429
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.328
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.484
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.355
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.193
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.492
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.552
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.422
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.599
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.487
Epoch: [5]  [  0/104]  eta: 0:03:07  lr: 0.000100  loss: 0.4193 (0.4193)  loss_classifier: 0.1600 (0.1600)  loss_box_reg: 0.2258 (0.2258)  loss_objectness: 0.0096 (0.0096)  loss_rpn_box_reg: 0.0240 (0.0240)  time: 1.8027  data: 0.7181  max mem: 6508
Epoch: [5]  [ 10/104]  eta: 0:01:46  lr: 0.000100  loss: 0.3269 (0.3370)  loss_classifier: 0.1390 (0.1268)  loss_box_reg: 0.1936 (0.1918)  loss_objectness: 0.0089 (0.0075)  loss_rpn_box_reg: 0.0080 (0.0109)  time: 1.1342  data: 0.0830  max mem: 6508
Epoch: [5]  [ 20/104]  eta: 0:01:33  lr: 0.000100  loss: 0.3544 (0.3734)  loss_classifier: 0.1277 (0.1388)  loss_box_reg: 0.2139 (0.2136)  loss_objectness: 0.0081 (0.0082)  loss_rpn_box_reg: 0.0104 (0.0128)  time: 1.0746  data: 0.0207  max mem: 6508
Epoch: [5]  [ 30/104]  eta: 0:01:20  lr: 0.000100  loss: 0.3695 (0.3627)  loss_classifier: 0.1277 (0.1345)  loss_box_reg: 0.2237 (0.2076)  loss_objectness: 0.0066 (0.0083)  loss_rpn_box_reg: 0.0114 (0.0123)  time: 1.0683  data: 0.0210  max mem: 6508
Epoch: [5]  [ 40/104]  eta: 0:01:09  lr: 0.000100  loss: 0.3503 (0.3576)  loss_classifier: 0.1212 (0.1295)  loss_box_reg: 0.2038 (0.2083)  loss_objectness: 0.0055 (0.0077)  loss_rpn_box_reg: 0.0103 (0.0121)  time: 1.0502  data: 0.0212  max mem: 6508
Epoch: [5]  [ 50/104]  eta: 0:00:57  lr: 0.000100  loss: 0.3338 (0.3440)  loss_classifier: 0.1079 (0.1239)  loss_box_reg: 0.1861 (0.2002)  loss_objectness: 0.0060 (0.0081)  loss_rpn_box_reg: 0.0099 (0.0118)  time: 1.0404  data: 0.0220  max mem: 6508
Epoch: [5]  [ 60/104]  eta: 0:00:46  lr: 0.000100  loss: 0.3008 (0.3420)  loss_classifier: 0.1079 (0.1222)  loss_box_reg: 0.1812 (0.2002)  loss_objectness: 0.0064 (0.0077)  loss_rpn_box_reg: 0.0101 (0.0118)  time: 1.0358  data: 0.0226  max mem: 6508
Epoch: [5]  [ 70/104]  eta: 0:00:36  lr: 0.000100  loss: 0.3433 (0.3416)  loss_classifier: 0.1082 (0.1228)  loss_box_reg: 0.1927 (0.2000)  loss_objectness: 0.0056 (0.0077)  loss_rpn_box_reg: 0.0085 (0.0111)  time: 1.0265  data: 0.0210  max mem: 6508
Epoch: [5]  [ 80/104]  eta: 0:00:25  lr: 0.000100  loss: 0.3123 (0.3360)  loss_classifier: 0.1012 (0.1207)  loss_box_reg: 0.1895 (0.1971)  loss_objectness: 0.0042 (0.0074)  loss_rpn_box_reg: 0.0065 (0.0108)  time: 1.0265  data: 0.0205  max mem: 6508
Epoch: [5]  [ 90/104]  eta: 0:00:14  lr: 0.000100  loss: 0.2546 (0.3282)  loss_classifier: 0.0837 (0.1173)  loss_box_reg: 0.1607 (0.1930)  loss_objectness: 0.0051 (0.0073)  loss_rpn_box_reg: 0.0070 (0.0105)  time: 1.0385  data: 0.0220  max mem: 6508
Epoch: [5]  [100/104]  eta: 0:00:04  lr: 0.000100  loss: 0.2826 (0.3304)  loss_classifier: 0.0950 (0.1175)  loss_box_reg: 0.1764 (0.1952)  loss_objectness: 0.0046 (0.0070)  loss_rpn_box_reg: 0.0077 (0.0108)  time: 1.0391  data: 0.0206  max mem: 6508
Epoch: [5]  [103/104]  eta: 0:00:01  lr: 0.000100  loss: 0.2938 (0.3298)  loss_classifier: 0.1015 (0.1174)  loss_box_reg: 0.1781 (0.1946)  loss_objectness: 0.0046 (0.0071)  loss_rpn_box_reg: 0.0076 (0.0107)  time: 1.0391  data: 0.0204  max mem: 6508
Epoch: [5] Total time: 0:01:49 (1.0535 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:34  model_time: 0.4812 (0.4812)  evaluator_time: 0.0558 (0.0558)  time: 1.3086  data: 0.7417  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4412 (0.4418)  evaluator_time: 0.0157 (0.0197)  time: 0.4870  data: 0.0209  max mem: 6508
Test: Total time: 0:00:13 (0.5221 s / it)
Averaged stats: model_time: 0.4412 (0.4418)  evaluator_time: 0.0157 (0.0197)
Accumulating evaluation results...
DONE (t=0.12s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.452
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.800
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.448
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.337
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.517
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.327
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.192
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.501
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.558
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.448
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.606
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.470
Epoch: [6]  [  0/104]  eta: 0:03:33  lr: 0.000010  loss: 0.4073 (0.4073)  loss_classifier: 0.1331 (0.1331)  loss_box_reg: 0.2401 (0.2401)  loss_objectness: 0.0098 (0.0098)  loss_rpn_box_reg: 0.0243 (0.0243)  time: 2.0575  data: 0.9893  max mem: 6508
Epoch: [6]  [ 10/104]  eta: 0:01:48  lr: 0.000010  loss: 0.2835 (0.3123)  loss_classifier: 0.1123 (0.1112)  loss_box_reg: 0.1738 (0.1859)  loss_objectness: 0.0055 (0.0062)  loss_rpn_box_reg: 0.0080 (0.0090)  time: 1.1551  data: 0.1058  max mem: 6508
Epoch: [6]  [ 20/104]  eta: 0:01:33  lr: 0.000010  loss: 0.2868 (0.3128)  loss_classifier: 0.1066 (0.1121)  loss_box_reg: 0.1738 (0.1840)  loss_objectness: 0.0049 (0.0067)  loss_rpn_box_reg: 0.0081 (0.0100)  time: 1.0704  data: 0.0195  max mem: 6508
Epoch: [6]  [ 30/104]  eta: 0:01:21  lr: 0.000010  loss: 0.2868 (0.3018)  loss_classifier: 0.0995 (0.1087)  loss_box_reg: 0.1600 (0.1769)  loss_objectness: 0.0046 (0.0066)  loss_rpn_box_reg: 0.0082 (0.0095)  time: 1.0642  data: 0.0202  max mem: 6508
Epoch: [6]  [ 40/104]  eta: 0:01:09  lr: 0.000010  loss: 0.2755 (0.3013)  loss_classifier: 0.0995 (0.1096)  loss_box_reg: 0.1557 (0.1758)  loss_objectness: 0.0041 (0.0062)  loss_rpn_box_reg: 0.0082 (0.0098)  time: 1.0481  data: 0.0203  max mem: 6508
Epoch: [6]  [ 50/104]  eta: 0:00:58  lr: 0.000010  loss: 0.2693 (0.2903)  loss_classifier: 0.0950 (0.1054)  loss_box_reg: 0.1473 (0.1697)  loss_objectness: 0.0033 (0.0057)  loss_rpn_box_reg: 0.0079 (0.0095)  time: 1.0435  data: 0.0249  max mem: 6508
Epoch: [6]  [ 60/104]  eta: 0:00:46  lr: 0.000010  loss: 0.2315 (0.2823)  loss_classifier: 0.0860 (0.1030)  loss_box_reg: 0.1282 (0.1646)  loss_objectness: 0.0033 (0.0055)  loss_rpn_box_reg: 0.0067 (0.0093)  time: 1.0364  data: 0.0246  max mem: 6508
Epoch: [6]  [ 70/104]  eta: 0:00:36  lr: 0.000010  loss: 0.2315 (0.2804)  loss_classifier: 0.0908 (0.1022)  loss_box_reg: 0.1212 (0.1636)  loss_objectness: 0.0036 (0.0052)  loss_rpn_box_reg: 0.0056 (0.0094)  time: 1.0283  data: 0.0202  max mem: 6508
Epoch: [6]  [ 80/104]  eta: 0:00:25  lr: 0.000010  loss: 0.2799 (0.2840)  loss_classifier: 0.0908 (0.1025)  loss_box_reg: 0.1810 (0.1670)  loss_objectness: 0.0036 (0.0052)  loss_rpn_box_reg: 0.0075 (0.0092)  time: 1.0330  data: 0.0210  max mem: 6508
Epoch: [6]  [ 90/104]  eta: 0:00:14  lr: 0.000010  loss: 0.2819 (0.2843)  loss_classifier: 0.1034 (0.1022)  loss_box_reg: 0.1826 (0.1673)  loss_objectness: 0.0040 (0.0053)  loss_rpn_box_reg: 0.0078 (0.0094)  time: 1.0453  data: 0.0224  max mem: 6508
Epoch: [6]  [100/104]  eta: 0:00:04  lr: 0.000010  loss: 0.2860 (0.2871)  loss_classifier: 0.1034 (0.1026)  loss_box_reg: 0.1680 (0.1699)  loss_objectness: 0.0045 (0.0052)  loss_rpn_box_reg: 0.0069 (0.0094)  time: 1.0449  data: 0.0208  max mem: 6508
Epoch: [6]  [103/104]  eta: 0:00:01  lr: 0.000010  loss: 0.2860 (0.2863)  loss_classifier: 0.1024 (0.1020)  loss_box_reg: 0.1680 (0.1698)  loss_objectness: 0.0034 (0.0052)  loss_rpn_box_reg: 0.0085 (0.0093)  time: 1.0419  data: 0.0200  max mem: 6508
Epoch: [6] Total time: 0:01:49 (1.0573 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:33  model_time: 0.5281 (0.5281)  evaluator_time: 0.0275 (0.0275)  time: 1.2963  data: 0.7170  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4374 (0.4399)  evaluator_time: 0.0158 (0.0183)  time: 0.4828  data: 0.0199  max mem: 6508
Test: Total time: 0:00:13 (0.5165 s / it)
Averaged stats: model_time: 0.4374 (0.4399)  evaluator_time: 0.0158 (0.0183)
Accumulating evaluation results...
DONE (t=0.11s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.478
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.821
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.482
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.332
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.539
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.367
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.208
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.518
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.576
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.444
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.623
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.492
Epoch: [7]  [  0/104]  eta: 0:03:22  lr: 0.000010  loss: 0.4399 (0.4399)  loss_classifier: 0.1619 (0.1619)  loss_box_reg: 0.2502 (0.2502)  loss_objectness: 0.0036 (0.0036)  loss_rpn_box_reg: 0.0242 (0.0242)  time: 1.9476  data: 0.8770  max mem: 6508
Epoch: [7]  [ 10/104]  eta: 0:01:47  lr: 0.000010  loss: 0.2756 (0.2936)  loss_classifier: 0.1017 (0.1059)  loss_box_reg: 0.1496 (0.1697)  loss_objectness: 0.0052 (0.0068)  loss_rpn_box_reg: 0.0107 (0.0112)  time: 1.1428  data: 0.0945  max mem: 6508
Epoch: [7]  [ 20/104]  eta: 0:01:33  lr: 0.000010  loss: 0.2791 (0.2945)  loss_classifier: 0.1017 (0.1067)  loss_box_reg: 0.1614 (0.1703)  loss_objectness: 0.0047 (0.0072)  loss_rpn_box_reg: 0.0082 (0.0103)  time: 1.0662  data: 0.0175  max mem: 6508
Epoch: [7]  [ 30/104]  eta: 0:01:20  lr: 0.000010  loss: 0.2799 (0.2795)  loss_classifier: 0.0968 (0.1006)  loss_box_reg: 0.1614 (0.1632)  loss_objectness: 0.0030 (0.0059)  loss_rpn_box_reg: 0.0076 (0.0098)  time: 1.0604  data: 0.0194  max mem: 6508
Epoch: [7]  [ 40/104]  eta: 0:01:09  lr: 0.000010  loss: 0.2400 (0.2744)  loss_classifier: 0.0840 (0.0997)  loss_box_reg: 0.1456 (0.1602)  loss_objectness: 0.0030 (0.0055)  loss_rpn_box_reg: 0.0067 (0.0090)  time: 1.0469  data: 0.0205  max mem: 6508
Epoch: [7]  [ 50/104]  eta: 0:00:57  lr: 0.000010  loss: 0.2466 (0.2673)  loss_classifier: 0.0838 (0.0960)  loss_box_reg: 0.1572 (0.1577)  loss_objectness: 0.0029 (0.0051)  loss_rpn_box_reg: 0.0061 (0.0085)  time: 1.0376  data: 0.0209  max mem: 6508
Epoch: [7]  [ 60/104]  eta: 0:00:46  lr: 0.000010  loss: 0.2689 (0.2686)  loss_classifier: 0.0875 (0.0953)  loss_box_reg: 0.1658 (0.1593)  loss_objectness: 0.0025 (0.0052)  loss_rpn_box_reg: 0.0078 (0.0088)  time: 1.0300  data: 0.0211  max mem: 6508
Epoch: [7]  [ 70/104]  eta: 0:00:35  lr: 0.000010  loss: 0.2628 (0.2667)  loss_classifier: 0.0850 (0.0944)  loss_box_reg: 0.1587 (0.1587)  loss_objectness: 0.0030 (0.0049)  loss_rpn_box_reg: 0.0078 (0.0087)  time: 1.0244  data: 0.0200  max mem: 6508
Epoch: [7]  [ 80/104]  eta: 0:00:25  lr: 0.000010  loss: 0.2451 (0.2705)  loss_classifier: 0.0906 (0.0968)  loss_box_reg: 0.1557 (0.1600)  loss_objectness: 0.0023 (0.0048)  loss_rpn_box_reg: 0.0069 (0.0090)  time: 1.0308  data: 0.0205  max mem: 6508
Epoch: [7]  [ 90/104]  eta: 0:00:14  lr: 0.000010  loss: 0.2572 (0.2687)  loss_classifier: 0.0913 (0.0959)  loss_box_reg: 0.1540 (0.1592)  loss_objectness: 0.0025 (0.0046)  loss_rpn_box_reg: 0.0075 (0.0090)  time: 1.0387  data: 0.0218  max mem: 6508
Epoch: [7]  [100/104]  eta: 0:00:04  lr: 0.000010  loss: 0.2582 (0.2735)  loss_classifier: 0.0950 (0.0978)  loss_box_reg: 0.1645 (0.1620)  loss_objectness: 0.0034 (0.0047)  loss_rpn_box_reg: 0.0078 (0.0091)  time: 1.0367  data: 0.0199  max mem: 6508
Epoch: [7]  [103/104]  eta: 0:00:01  lr: 0.000010  loss: 0.2582 (0.2739)  loss_classifier: 0.1008 (0.0976)  loss_box_reg: 0.1679 (0.1626)  loss_objectness: 0.0030 (0.0047)  loss_rpn_box_reg: 0.0075 (0.0090)  time: 1.0379  data: 0.0195  max mem: 6508
Epoch: [7] Total time: 0:01:49 (1.0520 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:32  model_time: 0.4987 (0.4987)  evaluator_time: 0.0271 (0.0271)  time: 1.2580  data: 0.7105  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4386 (0.4406)  evaluator_time: 0.0146 (0.0174)  time: 0.4851  data: 0.0209  max mem: 6508
Test: Total time: 0:00:13 (0.5171 s / it)
Averaged stats: model_time: 0.4386 (0.4406)  evaluator_time: 0.0146 (0.0174)
Accumulating evaluation results...
DONE (t=0.13s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.482
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.826
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.506
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.344
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.549
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.373
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.210
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.518
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.576
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.444
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.630
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.513
Epoch: [8]  [  0/104]  eta: 0:03:25  lr: 0.000010  loss: 0.3998 (0.3998)  loss_classifier: 0.1506 (0.1506)  loss_box_reg: 0.2245 (0.2245)  loss_objectness: 0.0096 (0.0096)  loss_rpn_box_reg: 0.0151 (0.0151)  time: 1.9714  data: 0.8839  max mem: 6508
Epoch: [8]  [ 10/104]  eta: 0:01:48  lr: 0.000010  loss: 0.3032 (0.2977)  loss_classifier: 0.1042 (0.1058)  loss_box_reg: 0.1803 (0.1755)  loss_objectness: 0.0038 (0.0049)  loss_rpn_box_reg: 0.0130 (0.0116)  time: 1.1560  data: 0.0993  max mem: 6508
Epoch: [8]  [ 20/104]  eta: 0:01:33  lr: 0.000010  loss: 0.2519 (0.2779)  loss_classifier: 0.0926 (0.0998)  loss_box_reg: 0.1590 (0.1635)  loss_objectness: 0.0033 (0.0044)  loss_rpn_box_reg: 0.0080 (0.0103)  time: 1.0757  data: 0.0202  max mem: 6508
Epoch: [8]  [ 30/104]  eta: 0:01:21  lr: 0.000010  loss: 0.2395 (0.2733)  loss_classifier: 0.0878 (0.0978)  loss_box_reg: 0.1522 (0.1611)  loss_objectness: 0.0033 (0.0044)  loss_rpn_box_reg: 0.0070 (0.0100)  time: 1.0648  data: 0.0192  max mem: 6508
Epoch: [8]  [ 40/104]  eta: 0:01:09  lr: 0.000010  loss: 0.2420 (0.2735)  loss_classifier: 0.0923 (0.0981)  loss_box_reg: 0.1517 (0.1608)  loss_objectness: 0.0044 (0.0047)  loss_rpn_box_reg: 0.0070 (0.0098)  time: 1.0438  data: 0.0193  max mem: 6508
Epoch: [8]  [ 50/104]  eta: 0:00:57  lr: 0.000010  loss: 0.2497 (0.2721)  loss_classifier: 0.0996 (0.0978)  loss_box_reg: 0.1467 (0.1600)  loss_objectness: 0.0039 (0.0049)  loss_rpn_box_reg: 0.0070 (0.0094)  time: 1.0306  data: 0.0204  max mem: 6508
Epoch: [8]  [ 60/104]  eta: 0:00:46  lr: 0.000010  loss: 0.2640 (0.2726)  loss_classifier: 0.0942 (0.0977)  loss_box_reg: 0.1487 (0.1609)  loss_objectness: 0.0035 (0.0047)  loss_rpn_box_reg: 0.0069 (0.0093)  time: 1.0261  data: 0.0208  max mem: 6508
Epoch: [8]  [ 70/104]  eta: 0:00:36  lr: 0.000010  loss: 0.2567 (0.2712)  loss_classifier: 0.0847 (0.0970)  loss_box_reg: 0.1487 (0.1604)  loss_objectness: 0.0038 (0.0046)  loss_rpn_box_reg: 0.0069 (0.0092)  time: 1.0291  data: 0.0220  max mem: 6508
Epoch: [8]  [ 80/104]  eta: 0:00:25  lr: 0.000010  loss: 0.2421 (0.2696)  loss_classifier: 0.0903 (0.0965)  loss_box_reg: 0.1393 (0.1595)  loss_objectness: 0.0039 (0.0046)  loss_rpn_box_reg: 0.0065 (0.0090)  time: 1.0323  data: 0.0223  max mem: 6508
Epoch: [8]  [ 90/104]  eta: 0:00:14  lr: 0.000010  loss: 0.2462 (0.2707)  loss_classifier: 0.0924 (0.0966)  loss_box_reg: 0.1409 (0.1606)  loss_objectness: 0.0037 (0.0045)  loss_rpn_box_reg: 0.0062 (0.0090)  time: 1.0354  data: 0.0213  max mem: 6508
Epoch: [8]  [100/104]  eta: 0:00:04  lr: 0.000010  loss: 0.2585 (0.2689)  loss_classifier: 0.0851 (0.0954)  loss_box_reg: 0.1527 (0.1602)  loss_objectness: 0.0033 (0.0044)  loss_rpn_box_reg: 0.0065 (0.0088)  time: 1.0377  data: 0.0201  max mem: 6508
Epoch: [8]  [103/104]  eta: 0:00:01  lr: 0.000010  loss: 0.2463 (0.2687)  loss_classifier: 0.0792 (0.0952)  loss_box_reg: 0.1509 (0.1602)  loss_objectness: 0.0033 (0.0044)  loss_rpn_box_reg: 0.0062 (0.0089)  time: 1.0379  data: 0.0197  max mem: 6508
Epoch: [8] Total time: 0:01:49 (1.0532 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:32  model_time: 0.5196 (0.5196)  evaluator_time: 0.0677 (0.0677)  time: 1.2434  data: 0.6343  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4391 (0.4423)  evaluator_time: 0.0157 (0.0205)  time: 0.4859  data: 0.0210  max mem: 6508
Test: Total time: 0:00:13 (0.5185 s / it)
Averaged stats: model_time: 0.4391 (0.4423)  evaluator_time: 0.0157 (0.0205)
Accumulating evaluation results...
DONE (t=0.10s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.483
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.828
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.492
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.335
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.545
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.386
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.210
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.522
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.576
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.445
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.629
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.528
Epoch: [9]  [  0/104]  eta: 0:03:23  lr: 0.000001  loss: 0.1893 (0.1893)  loss_classifier: 0.0680 (0.0680)  loss_box_reg: 0.1151 (0.1151)  loss_objectness: 0.0013 (0.0013)  loss_rpn_box_reg: 0.0050 (0.0050)  time: 1.9530  data: 0.8919  max mem: 6508
Epoch: [9]  [ 10/104]  eta: 0:01:48  lr: 0.000001  loss: 0.2888 (0.2850)  loss_classifier: 0.1062 (0.1099)  loss_box_reg: 0.1607 (0.1611)  loss_objectness: 0.0035 (0.0051)  loss_rpn_box_reg: 0.0061 (0.0089)  time: 1.1501  data: 0.0951  max mem: 6508
Epoch: [9]  [ 20/104]  eta: 0:01:33  lr: 0.000001  loss: 0.2615 (0.2629)  loss_classifier: 0.0999 (0.0993)  loss_box_reg: 0.1388 (0.1513)  loss_objectness: 0.0040 (0.0046)  loss_rpn_box_reg: 0.0060 (0.0077)  time: 1.0771  data: 0.0198  max mem: 6508
Epoch: [9]  [ 30/104]  eta: 0:01:21  lr: 0.000001  loss: 0.2615 (0.2642)  loss_classifier: 0.0915 (0.0972)  loss_box_reg: 0.1451 (0.1539)  loss_objectness: 0.0042 (0.0046)  loss_rpn_box_reg: 0.0081 (0.0085)  time: 1.0698  data: 0.0227  max mem: 6508
Epoch: [9]  [ 40/104]  eta: 0:01:09  lr: 0.000001  loss: 0.2488 (0.2595)  loss_classifier: 0.0905 (0.0954)  loss_box_reg: 0.1451 (0.1510)  loss_objectness: 0.0039 (0.0045)  loss_rpn_box_reg: 0.0103 (0.0086)  time: 1.0462  data: 0.0204  max mem: 6508
Epoch: [9]  [ 50/104]  eta: 0:00:58  lr: 0.000001  loss: 0.2303 (0.2553)  loss_classifier: 0.0789 (0.0925)  loss_box_reg: 0.1414 (0.1498)  loss_objectness: 0.0034 (0.0043)  loss_rpn_box_reg: 0.0077 (0.0087)  time: 1.0397  data: 0.0205  max mem: 6508
Epoch: [9]  [ 60/104]  eta: 0:00:46  lr: 0.000001  loss: 0.2391 (0.2557)  loss_classifier: 0.0853 (0.0925)  loss_box_reg: 0.1478 (0.1502)  loss_objectness: 0.0037 (0.0043)  loss_rpn_box_reg: 0.0073 (0.0088)  time: 1.0324  data: 0.0208  max mem: 6508
Epoch: [9]  [ 70/104]  eta: 0:00:36  lr: 0.000001  loss: 0.2392 (0.2576)  loss_classifier: 0.0898 (0.0927)  loss_box_reg: 0.1430 (0.1518)  loss_objectness: 0.0033 (0.0042)  loss_rpn_box_reg: 0.0072 (0.0089)  time: 1.0274  data: 0.0208  max mem: 6508
Epoch: [9]  [ 80/104]  eta: 0:00:25  lr: 0.000001  loss: 0.2606 (0.2628)  loss_classifier: 0.0934 (0.0937)  loss_box_reg: 0.1607 (0.1561)  loss_objectness: 0.0022 (0.0040)  loss_rpn_box_reg: 0.0072 (0.0090)  time: 1.0352  data: 0.0209  max mem: 6508
Epoch: [9]  [ 90/104]  eta: 0:00:14  lr: 0.000001  loss: 0.2535 (0.2627)  loss_classifier: 0.0889 (0.0935)  loss_box_reg: 0.1581 (0.1562)  loss_objectness: 0.0030 (0.0041)  loss_rpn_box_reg: 0.0065 (0.0090)  time: 1.0406  data: 0.0204  max mem: 6508
Epoch: [9]  [100/104]  eta: 0:00:04  lr: 0.000001  loss: 0.2500 (0.2615)  loss_classifier: 0.0813 (0.0928)  loss_box_reg: 0.1495 (0.1559)  loss_objectness: 0.0036 (0.0040)  loss_rpn_box_reg: 0.0060 (0.0087)  time: 1.0400  data: 0.0197  max mem: 6508
Epoch: [9]  [103/104]  eta: 0:00:01  lr: 0.000001  loss: 0.2444 (0.2629)  loss_classifier: 0.0784 (0.0932)  loss_box_reg: 0.1495 (0.1569)  loss_objectness: 0.0037 (0.0041)  loss_rpn_box_reg: 0.0062 (0.0087)  time: 1.0406  data: 0.0196  max mem: 6508
Epoch: [9] Total time: 0:01:49 (1.0562 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:32  model_time: 0.4913 (0.4913)  evaluator_time: 0.0282 (0.0282)  time: 1.2392  data: 0.6994  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4355 (0.4357)  evaluator_time: 0.0149 (0.0173)  time: 0.4806  data: 0.0211  max mem: 6508
Test: Total time: 0:00:13 (0.5108 s / it)
Averaged stats: model_time: 0.4355 (0.4357)  evaluator_time: 0.0149 (0.0173)
Accumulating evaluation results...
DONE (t=0.11s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.485
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.830
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.489
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.345
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.545
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.383
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.210
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.519
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.574
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.447
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.622
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.518
Epoch: [10]  [  0/104]  eta: 0:03:18  lr: 0.000001  loss: 0.3089 (0.3089)  loss_classifier: 0.1102 (0.1102)  loss_box_reg: 0.1835 (0.1835)  loss_objectness: 0.0038 (0.0038)  loss_rpn_box_reg: 0.0114 (0.0114)  time: 1.9111  data: 0.8486  max mem: 6508
Epoch: [10]  [ 10/104]  eta: 0:01:47  lr: 0.000001  loss: 0.2657 (0.2665)  loss_classifier: 0.0903 (0.0977)  loss_box_reg: 0.1476 (0.1562)  loss_objectness: 0.0035 (0.0042)  loss_rpn_box_reg: 0.0075 (0.0084)  time: 1.1399  data: 0.0912  max mem: 6508
Epoch: [10]  [ 20/104]  eta: 0:01:35  lr: 0.000001  loss: 0.2605 (0.2809)  loss_classifier: 0.0932 (0.1030)  loss_box_reg: 0.1476 (0.1628)  loss_objectness: 0.0034 (0.0059)  loss_rpn_box_reg: 0.0076 (0.0092)  time: 1.0934  data: 0.0224  max mem: 6508
Epoch: [10]  [ 30/104]  eta: 0:01:25  lr: 0.000001  loss: 0.2682 (0.2800)  loss_classifier: 0.0955 (0.1011)  loss_box_reg: 0.1592 (0.1640)  loss_objectness: 0.0035 (0.0051)  loss_rpn_box_reg: 0.0078 (0.0098)  time: 1.1623  data: 0.0358  max mem: 6508
Epoch: [10]  [ 40/104]  eta: 0:01:12  lr: 0.000001  loss: 0.2551 (0.2719)  loss_classifier: 0.0816 (0.0969)  loss_box_reg: 0.1583 (0.1608)  loss_objectness: 0.0033 (0.0047)  loss_rpn_box_reg: 0.0069 (0.0094)  time: 1.1410  data: 0.0376  max mem: 6508
Epoch: [10]  [ 50/104]  eta: 0:01:00  lr: 0.000001  loss: 0.2229 (0.2650)  loss_classifier: 0.0771 (0.0944)  loss_box_reg: 0.1324 (0.1573)  loss_objectness: 0.0032 (0.0044)  loss_rpn_box_reg: 0.0054 (0.0089)  time: 1.0621  data: 0.0288  max mem: 6508
Epoch: [10]  [ 60/104]  eta: 0:00:48  lr: 0.000001  loss: 0.2212 (0.2611)  loss_classifier: 0.0734 (0.0934)  loss_box_reg: 0.1324 (0.1546)  loss_objectness: 0.0029 (0.0043)  loss_rpn_box_reg: 0.0052 (0.0088)  time: 1.0553  data: 0.0267  max mem: 6508
Epoch: [10]  [ 70/104]  eta: 0:00:37  lr: 0.000001  loss: 0.2255 (0.2605)  loss_classifier: 0.0734 (0.0928)  loss_box_reg: 0.1316 (0.1547)  loss_objectness: 0.0038 (0.0042)  loss_rpn_box_reg: 0.0060 (0.0087)  time: 1.0582  data: 0.0265  max mem: 6508
Epoch: [10]  [ 80/104]  eta: 0:00:26  lr: 0.000001  loss: 0.2585 (0.2621)  loss_classifier: 0.0922 (0.0928)  loss_box_reg: 0.1622 (0.1563)  loss_objectness: 0.0034 (0.0041)  loss_rpn_box_reg: 0.0080 (0.0089)  time: 1.0439  data: 0.0230  max mem: 6508
Epoch: [10]  [ 90/104]  eta: 0:00:15  lr: 0.000001  loss: 0.2750 (0.2630)  loss_classifier: 0.0931 (0.0927)  loss_box_reg: 0.1622 (0.1573)  loss_objectness: 0.0034 (0.0042)  loss_rpn_box_reg: 0.0080 (0.0089)  time: 1.0459  data: 0.0230  max mem: 6508
Epoch: [10]  [100/104]  eta: 0:00:04  lr: 0.000001  loss: 0.2316 (0.2620)  loss_classifier: 0.0895 (0.0925)  loss_box_reg: 0.1439 (0.1569)  loss_objectness: 0.0026 (0.0040)  loss_rpn_box_reg: 0.0062 (0.0087)  time: 1.0466  data: 0.0214  max mem: 6508
Epoch: [10]  [103/104]  eta: 0:00:01  lr: 0.000001  loss: 0.2247 (0.2615)  loss_classifier: 0.0815 (0.0924)  loss_box_reg: 0.1439 (0.1564)  loss_objectness: 0.0025 (0.0040)  loss_rpn_box_reg: 0.0065 (0.0087)  time: 1.0447  data: 0.0206  max mem: 6508
Epoch: [10] Total time: 0:01:52 (1.0842 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:33  model_time: 0.4919 (0.4919)  evaluator_time: 0.0472 (0.0472)  time: 1.2793  data: 0.7246  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4423 (0.4451)  evaluator_time: 0.0137 (0.0207)  time: 0.4948  data: 0.0227  max mem: 6508
Test: Total time: 0:00:13 (0.5284 s / it)
Averaged stats: model_time: 0.4423 (0.4451)  evaluator_time: 0.0137 (0.0207)
Accumulating evaluation results...
DONE (t=0.12s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.488
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.833
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.504
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.346
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.548
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.393
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.213
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.523
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.578
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.447
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.626
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.532
Epoch: [11]  [  0/104]  eta: 0:03:19  lr: 0.000001  loss: 0.2695 (0.2695)  loss_classifier: 0.0898 (0.0898)  loss_box_reg: 0.1705 (0.1705)  loss_objectness: 0.0016 (0.0016)  loss_rpn_box_reg: 0.0075 (0.0075)  time: 1.9135  data: 0.7632  max mem: 6508
Epoch: [11]  [ 10/104]  eta: 0:01:48  lr: 0.000001  loss: 0.2695 (0.2629)  loss_classifier: 0.0898 (0.0921)  loss_box_reg: 0.1633 (0.1565)  loss_objectness: 0.0020 (0.0031)  loss_rpn_box_reg: 0.0095 (0.0111)  time: 1.1551  data: 0.0876  max mem: 6508
Epoch: [11]  [ 20/104]  eta: 0:01:33  lr: 0.000001  loss: 0.2545 (0.2668)  loss_classifier: 0.0900 (0.0933)  loss_box_reg: 0.1567 (0.1594)  loss_objectness: 0.0022 (0.0033)  loss_rpn_box_reg: 0.0080 (0.0107)  time: 1.0751  data: 0.0213  max mem: 6508
Epoch: [11]  [ 30/104]  eta: 0:01:21  lr: 0.000001  loss: 0.2302 (0.2587)  loss_classifier: 0.0824 (0.0928)  loss_box_reg: 0.1386 (0.1530)  loss_objectness: 0.0025 (0.0035)  loss_rpn_box_reg: 0.0062 (0.0095)  time: 1.0620  data: 0.0225  max mem: 6508
Epoch: [11]  [ 40/104]  eta: 0:01:09  lr: 0.000001  loss: 0.2295 (0.2561)  loss_classifier: 0.0814 (0.0911)  loss_box_reg: 0.1414 (0.1524)  loss_objectness: 0.0024 (0.0033)  loss_rpn_box_reg: 0.0055 (0.0093)  time: 1.0423  data: 0.0211  max mem: 6508
Epoch: [11]  [ 50/104]  eta: 0:00:57  lr: 0.000001  loss: 0.2392 (0.2601)  loss_classifier: 0.0881 (0.0922)  loss_box_reg: 0.1480 (0.1548)  loss_objectness: 0.0032 (0.0037)  loss_rpn_box_reg: 0.0062 (0.0094)  time: 1.0284  data: 0.0201  max mem: 6508
Epoch: [11]  [ 60/104]  eta: 0:00:46  lr: 0.000001  loss: 0.2893 (0.2606)  loss_classifier: 0.0993 (0.0921)  loss_box_reg: 0.1679 (0.1556)  loss_objectness: 0.0031 (0.0037)  loss_rpn_box_reg: 0.0080 (0.0092)  time: 1.0265  data: 0.0197  max mem: 6508
Epoch: [11]  [ 70/104]  eta: 0:00:35  lr: 0.000001  loss: 0.2819 (0.2620)  loss_classifier: 0.0993 (0.0922)  loss_box_reg: 0.1679 (0.1571)  loss_objectness: 0.0029 (0.0037)  loss_rpn_box_reg: 0.0071 (0.0089)  time: 1.0316  data: 0.0195  max mem: 6508
Epoch: [11]  [ 80/104]  eta: 0:00:25  lr: 0.000001  loss: 0.2583 (0.2619)  loss_classifier: 0.0874 (0.0921)  loss_box_reg: 0.1529 (0.1574)  loss_objectness: 0.0028 (0.0037)  loss_rpn_box_reg: 0.0071 (0.0087)  time: 1.0343  data: 0.0199  max mem: 6508
Epoch: [11]  [ 90/104]  eta: 0:00:14  lr: 0.000001  loss: 0.2398 (0.2594)  loss_classifier: 0.0854 (0.0916)  loss_box_reg: 0.1406 (0.1555)  loss_objectness: 0.0026 (0.0037)  loss_rpn_box_reg: 0.0066 (0.0086)  time: 1.0461  data: 0.0237  max mem: 6508
Epoch: [11]  [100/104]  eta: 0:00:04  lr: 0.000001  loss: 0.2494 (0.2586)  loss_classifier: 0.0855 (0.0914)  loss_box_reg: 0.1472 (0.1548)  loss_objectness: 0.0029 (0.0037)  loss_rpn_box_reg: 0.0070 (0.0086)  time: 1.0521  data: 0.0247  max mem: 6508
Epoch: [11]  [103/104]  eta: 0:00:01  lr: 0.000001  loss: 0.2512 (0.2602)  loss_classifier: 0.0885 (0.0920)  loss_box_reg: 0.1508 (0.1557)  loss_objectness: 0.0036 (0.0039)  loss_rpn_box_reg: 0.0072 (0.0087)  time: 1.0448  data: 0.0215  max mem: 6508
Epoch: [11] Total time: 0:01:49 (1.0555 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:33  model_time: 0.5373 (0.5373)  evaluator_time: 0.0450 (0.0450)  time: 1.2868  data: 0.6877  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4405 (0.4423)  evaluator_time: 0.0152 (0.0187)  time: 0.4872  data: 0.0222  max mem: 6508
Test: Total time: 0:00:13 (0.5198 s / it)
Averaged stats: model_time: 0.4405 (0.4423)  evaluator_time: 0.0152 (0.0187)
Accumulating evaluation results...
DONE (t=0.11s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.490
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.833
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.511
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.346
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.551
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.396
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.215
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.524
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.579
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.448
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.628
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.523
Epoch: [12]  [  0/104]  eta: 0:03:19  lr: 0.000000  loss: 0.3497 (0.3497)  loss_classifier: 0.1162 (0.1162)  loss_box_reg: 0.2069 (0.2069)  loss_objectness: 0.0060 (0.0060)  loss_rpn_box_reg: 0.0205 (0.0205)  time: 1.9170  data: 0.8726  max mem: 6508
Epoch: [12]  [ 10/104]  eta: 0:01:47  lr: 0.000000  loss: 0.2639 (0.2456)  loss_classifier: 0.0837 (0.0876)  loss_box_reg: 0.1679 (0.1477)  loss_objectness: 0.0025 (0.0031)  loss_rpn_box_reg: 0.0052 (0.0071)  time: 1.1453  data: 0.0969  max mem: 6508
Epoch: [12]  [ 20/104]  eta: 0:01:33  lr: 0.000000  loss: 0.2498 (0.2473)  loss_classifier: 0.0865 (0.0913)  loss_box_reg: 0.1379 (0.1448)  loss_objectness: 0.0034 (0.0034)  loss_rpn_box_reg: 0.0062 (0.0078)  time: 1.0765  data: 0.0202  max mem: 6508
Epoch: [12]  [ 30/104]  eta: 0:01:21  lr: 0.000000  loss: 0.2490 (0.2526)  loss_classifier: 0.0865 (0.0924)  loss_box_reg: 0.1386 (0.1474)  loss_objectness: 0.0047 (0.0044)  loss_rpn_box_reg: 0.0075 (0.0084)  time: 1.0742  data: 0.0201  max mem: 6508
Epoch: [12]  [ 40/104]  eta: 0:01:09  lr: 0.000000  loss: 0.2490 (0.2493)  loss_classifier: 0.0856 (0.0914)  loss_box_reg: 0.1379 (0.1453)  loss_objectness: 0.0039 (0.0043)  loss_rpn_box_reg: 0.0075 (0.0083)  time: 1.0522  data: 0.0191  max mem: 6508
Epoch: [12]  [ 50/104]  eta: 0:00:58  lr: 0.000000  loss: 0.2500 (0.2548)  loss_classifier: 0.0856 (0.0921)  loss_box_reg: 0.1310 (0.1499)  loss_objectness: 0.0031 (0.0042)  loss_rpn_box_reg: 0.0074 (0.0086)  time: 1.0406  data: 0.0189  max mem: 6508
Epoch: [12]  [ 60/104]  eta: 0:00:46  lr: 0.000000  loss: 0.3066 (0.2637)  loss_classifier: 0.0943 (0.0945)  loss_box_reg: 0.1809 (0.1558)  loss_objectness: 0.0040 (0.0043)  loss_rpn_box_reg: 0.0091 (0.0090)  time: 1.0306  data: 0.0201  max mem: 6508
Epoch: [12]  [ 70/104]  eta: 0:00:36  lr: 0.000000  loss: 0.2656 (0.2608)  loss_classifier: 0.0877 (0.0930)  loss_box_reg: 0.1641 (0.1547)  loss_objectness: 0.0038 (0.0043)  loss_rpn_box_reg: 0.0085 (0.0088)  time: 1.0225  data: 0.0218  max mem: 6508
Epoch: [12]  [ 80/104]  eta: 0:00:25  lr: 0.000000  loss: 0.2258 (0.2579)  loss_classifier: 0.0735 (0.0922)  loss_box_reg: 0.1359 (0.1531)  loss_objectness: 0.0026 (0.0041)  loss_rpn_box_reg: 0.0066 (0.0086)  time: 1.0465  data: 0.0231  max mem: 6508
Epoch: [12]  [ 90/104]  eta: 0:00:14  lr: 0.000000  loss: 0.2576 (0.2613)  loss_classifier: 0.0903 (0.0929)  loss_box_reg: 0.1535 (0.1554)  loss_objectness: 0.0026 (0.0042)  loss_rpn_box_reg: 0.0069 (0.0089)  time: 1.0554  data: 0.0229  max mem: 6508
Epoch: [12]  [100/104]  eta: 0:00:04  lr: 0.000000  loss: 0.2593 (0.2589)  loss_classifier: 0.0903 (0.0915)  loss_box_reg: 0.1555 (0.1548)  loss_objectness: 0.0030 (0.0041)  loss_rpn_box_reg: 0.0068 (0.0086)  time: 1.0412  data: 0.0205  max mem: 6508
Epoch: [12]  [103/104]  eta: 0:00:01  lr: 0.000000  loss: 0.2605 (0.2601)  loss_classifier: 0.0918 (0.0920)  loss_box_reg: 0.1555 (0.1553)  loss_objectness: 0.0030 (0.0041)  loss_rpn_box_reg: 0.0065 (0.0087)  time: 1.0399  data: 0.0198  max mem: 6508
Epoch: [12] Total time: 0:01:50 (1.0596 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:40  model_time: 0.4978 (0.4978)  evaluator_time: 0.0399 (0.0399)  time: 1.5401  data: 0.9932  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4383 (0.4400)  evaluator_time: 0.0135 (0.0174)  time: 0.4859  data: 0.0217  max mem: 6508
Test: Total time: 0:00:13 (0.5286 s / it)
Averaged stats: model_time: 0.4383 (0.4400)  evaluator_time: 0.0135 (0.0174)
Accumulating evaluation results...
DONE (t=0.11s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.491
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.833
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.514
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.347
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.551
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.396
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.216
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.524
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.580
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.448
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.629
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.523
Epoch: [13]  [  0/104]  eta: 0:03:24  lr: 0.000000  loss: 0.2393 (0.2393)  loss_classifier: 0.0915 (0.0915)  loss_box_reg: 0.1408 (0.1408)  loss_objectness: 0.0010 (0.0010)  loss_rpn_box_reg: 0.0061 (0.0061)  time: 1.9701  data: 0.8935  max mem: 6508
Epoch: [13]  [ 10/104]  eta: 0:01:48  lr: 0.000000  loss: 0.2910 (0.2708)  loss_classifier: 0.1009 (0.0982)  loss_box_reg: 0.1593 (0.1614)  loss_objectness: 0.0021 (0.0034)  loss_rpn_box_reg: 0.0082 (0.0078)  time: 1.1562  data: 0.1002  max mem: 6508
Epoch: [13]  [ 20/104]  eta: 0:01:34  lr: 0.000000  loss: 0.2833 (0.2792)  loss_classifier: 0.1029 (0.1013)  loss_box_reg: 0.1593 (0.1650)  loss_objectness: 0.0035 (0.0041)  loss_rpn_box_reg: 0.0085 (0.0089)  time: 1.0812  data: 0.0229  max mem: 6508
Epoch: [13]  [ 30/104]  eta: 0:01:21  lr: 0.000000  loss: 0.2602 (0.2660)  loss_classifier: 0.0801 (0.0950)  loss_box_reg: 0.1569 (0.1582)  loss_objectness: 0.0033 (0.0040)  loss_rpn_box_reg: 0.0079 (0.0089)  time: 1.0702  data: 0.0222  max mem: 6508
Epoch: [13]  [ 40/104]  eta: 0:01:09  lr: 0.000000  loss: 0.2506 (0.2652)  loss_classifier: 0.0801 (0.0937)  loss_box_reg: 0.1573 (0.1587)  loss_objectness: 0.0028 (0.0041)  loss_rpn_box_reg: 0.0074 (0.0087)  time: 1.0433  data: 0.0199  max mem: 6508
Epoch: [13]  [ 50/104]  eta: 0:00:57  lr: 0.000000  loss: 0.2642 (0.2710)  loss_classifier: 0.0905 (0.0958)  loss_box_reg: 0.1610 (0.1622)  loss_objectness: 0.0033 (0.0039)  loss_rpn_box_reg: 0.0074 (0.0091)  time: 1.0326  data: 0.0208  max mem: 6508
Epoch: [13]  [ 60/104]  eta: 0:00:46  lr: 0.000000  loss: 0.2453 (0.2643)  loss_classifier: 0.0866 (0.0942)  loss_box_reg: 0.1528 (0.1572)  loss_objectness: 0.0037 (0.0041)  loss_rpn_box_reg: 0.0067 (0.0089)  time: 1.0309  data: 0.0213  max mem: 6508
Epoch: [13]  [ 70/104]  eta: 0:00:36  lr: 0.000000  loss: 0.2453 (0.2652)  loss_classifier: 0.0866 (0.0948)  loss_box_reg: 0.1528 (0.1573)  loss_objectness: 0.0042 (0.0041)  loss_rpn_box_reg: 0.0079 (0.0090)  time: 1.0320  data: 0.0206  max mem: 6508
Epoch: [13]  [ 80/104]  eta: 0:00:25  lr: 0.000000  loss: 0.2683 (0.2626)  loss_classifier: 0.0866 (0.0937)  loss_box_reg: 0.1559 (0.1561)  loss_objectness: 0.0035 (0.0041)  loss_rpn_box_reg: 0.0085 (0.0087)  time: 1.0314  data: 0.0197  max mem: 6508
Epoch: [13]  [ 90/104]  eta: 0:00:14  lr: 0.000000  loss: 0.2463 (0.2607)  loss_classifier: 0.0811 (0.0925)  loss_box_reg: 0.1417 (0.1554)  loss_objectness: 0.0035 (0.0041)  loss_rpn_box_reg: 0.0064 (0.0087)  time: 1.0334  data: 0.0196  max mem: 6508
Epoch: [13]  [100/104]  eta: 0:00:04  lr: 0.000000  loss: 0.2353 (0.2590)  loss_classifier: 0.0811 (0.0918)  loss_box_reg: 0.1483 (0.1546)  loss_objectness: 0.0020 (0.0039)  loss_rpn_box_reg: 0.0069 (0.0086)  time: 1.0410  data: 0.0193  max mem: 6508
Epoch: [13]  [103/104]  eta: 0:00:01  lr: 0.000000  loss: 0.2353 (0.2591)  loss_classifier: 0.0850 (0.0920)  loss_box_reg: 0.1483 (0.1545)  loss_objectness: 0.0021 (0.0039)  loss_rpn_box_reg: 0.0070 (0.0086)  time: 1.0392  data: 0.0185  max mem: 6508
Epoch: [13] Total time: 0:01:49 (1.0554 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:34  model_time: 0.5242 (0.5242)  evaluator_time: 0.0259 (0.0259)  time: 1.3083  data: 0.7425  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4377 (0.4397)  evaluator_time: 0.0137 (0.0181)  time: 0.4854  data: 0.0227  max mem: 6508
Test: Total time: 0:00:13 (0.5196 s / it)
Averaged stats: model_time: 0.4377 (0.4397)  evaluator_time: 0.0137 (0.0181)
Accumulating evaluation results...
DONE (t=0.13s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.491
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.833
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.514
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.347
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.551
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.396
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.216
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.524
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.580
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.448
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.629
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.523
Epoch: [14]  [  0/104]  eta: 0:03:12  lr: 0.000000  loss: 0.3410 (0.3410)  loss_classifier: 0.1312 (0.1312)  loss_box_reg: 0.1974 (0.1974)  loss_objectness: 0.0029 (0.0029)  loss_rpn_box_reg: 0.0096 (0.0096)  time: 1.8523  data: 0.7088  max mem: 6508
Epoch: [14]  [ 10/104]  eta: 0:01:46  lr: 0.000000  loss: 0.2914 (0.2693)  loss_classifier: 0.1028 (0.0986)  loss_box_reg: 0.1671 (0.1594)  loss_objectness: 0.0027 (0.0031)  loss_rpn_box_reg: 0.0082 (0.0081)  time: 1.1355  data: 0.0833  max mem: 6508
Epoch: [14]  [ 20/104]  eta: 0:01:33  lr: 0.000000  loss: 0.2498 (0.2603)  loss_classifier: 0.0872 (0.0934)  loss_box_reg: 0.1375 (0.1540)  loss_objectness: 0.0027 (0.0036)  loss_rpn_box_reg: 0.0078 (0.0092)  time: 1.0747  data: 0.0225  max mem: 6508
Epoch: [14]  [ 30/104]  eta: 0:01:20  lr: 0.000000  loss: 0.2601 (0.2608)  loss_classifier: 0.0880 (0.0931)  loss_box_reg: 0.1574 (0.1557)  loss_objectness: 0.0030 (0.0036)  loss_rpn_box_reg: 0.0063 (0.0085)  time: 1.0702  data: 0.0222  max mem: 6508
Epoch: [14]  [ 40/104]  eta: 0:01:09  lr: 0.000000  loss: 0.2729 (0.2664)  loss_classifier: 0.0955 (0.0959)  loss_box_reg: 0.1624 (0.1577)  loss_objectness: 0.0032 (0.0043)  loss_rpn_box_reg: 0.0082 (0.0085)  time: 1.0459  data: 0.0205  max mem: 6508
Epoch: [14]  [ 50/104]  eta: 0:00:57  lr: 0.000000  loss: 0.2713 (0.2574)  loss_classifier: 0.0907 (0.0935)  loss_box_reg: 0.1412 (0.1514)  loss_objectness: 0.0025 (0.0041)  loss_rpn_box_reg: 0.0082 (0.0083)  time: 1.0353  data: 0.0213  max mem: 6508
Epoch: [14]  [ 60/104]  eta: 0:00:46  lr: 0.000000  loss: 0.2197 (0.2578)  loss_classifier: 0.0854 (0.0934)  loss_box_reg: 0.1366 (0.1518)  loss_objectness: 0.0021 (0.0039)  loss_rpn_box_reg: 0.0065 (0.0087)  time: 1.0321  data: 0.0218  max mem: 6508
Epoch: [14]  [ 70/104]  eta: 0:00:36  lr: 0.000000  loss: 0.2562 (0.2615)  loss_classifier: 0.0945 (0.0944)  loss_box_reg: 0.1455 (0.1542)  loss_objectness: 0.0028 (0.0041)  loss_rpn_box_reg: 0.0077 (0.0088)  time: 1.0295  data: 0.0213  max mem: 6508
Epoch: [14]  [ 80/104]  eta: 0:00:25  lr: 0.000000  loss: 0.2723 (0.2631)  loss_classifier: 0.0945 (0.0939)  loss_box_reg: 0.1650 (0.1559)  loss_objectness: 0.0033 (0.0041)  loss_rpn_box_reg: 0.0102 (0.0092)  time: 1.0303  data: 0.0202  max mem: 6508
Epoch: [14]  [ 90/104]  eta: 0:00:14  lr: 0.000000  loss: 0.2576 (0.2605)  loss_classifier: 0.0889 (0.0929)  loss_box_reg: 0.1560 (0.1547)  loss_objectness: 0.0022 (0.0040)  loss_rpn_box_reg: 0.0074 (0.0089)  time: 1.0369  data: 0.0208  max mem: 6508
Epoch: [14]  [100/104]  eta: 0:00:04  lr: 0.000000  loss: 0.2431 (0.2599)  loss_classifier: 0.0874 (0.0921)  loss_box_reg: 0.1418 (0.1552)  loss_objectness: 0.0022 (0.0039)  loss_rpn_box_reg: 0.0055 (0.0087)  time: 1.0391  data: 0.0203  max mem: 6508
Epoch: [14]  [103/104]  eta: 0:00:01  lr: 0.000000  loss: 0.2379 (0.2594)  loss_classifier: 0.0836 (0.0920)  loss_box_reg: 0.1418 (0.1549)  loss_objectness: 0.0022 (0.0039)  loss_rpn_box_reg: 0.0055 (0.0086)  time: 1.0370  data: 0.0194  max mem: 6508
Epoch: [14] Total time: 0:01:49 (1.0532 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:31  model_time: 0.5288 (0.5288)  evaluator_time: 0.0853 (0.0853)  time: 1.2276  data: 0.5798  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4429 (0.4433)  evaluator_time: 0.0207 (0.0252)  time: 0.4958  data: 0.0231  max mem: 6508
Test: Total time: 0:00:13 (0.5262 s / it)
Averaged stats: model_time: 0.4429 (0.4433)  evaluator_time: 0.0207 (0.0252)
Accumulating evaluation results...
DONE (t=0.11s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.491
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.833
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.514
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.347
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.551
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.396
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.216
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.524
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.580
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.448
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.629
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.523

#save adamn
import pickle
Filename = "FRCNN2adamn.pkl"
# Define the file path where you want to save the model
filename = "/content/drive/MyDrive/dataset1/FRCNN2adamn.pkl"

# Save the model to the specified file path
torch.save(model.state_dict(), filename)
# Save the Modle to file in the current working directory
with open(Filename, 'wb') as file:
    pickle.dump(model, file)
# Load the Model back from file
with open(Filename, 'rb') as file:
    model = pickle.load(file)
model

FasterRCNN(
  (transform): GeneralizedRCNNTransform(
      Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
      Resize(min_size=(800,), max_size=1333, mode='bilinear')
  )
  (backbone): BackboneWithFPN(
    (body): IntermediateLayerGetter(
      (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)
      (bn1): FrozenBatchNorm2d(64, eps=0.0)
      (relu): ReLU(inplace=True)
      (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)
      (layer1): Sequential(
        (0): Bottleneck(
          (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(64, eps=0.0)
          (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(64, eps=0.0)
          (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(256, eps=0.0)
          (relu): ReLU(inplace=True)
          (downsample): Sequential(
            (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
            (1): FrozenBatchNorm2d(256, eps=0.0)
          )
        )
        (1): Bottleneck(
          (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(64, eps=0.0)
          (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(64, eps=0.0)
          (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(256, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (2): Bottleneck(
          (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(64, eps=0.0)
          (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(64, eps=0.0)
          (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(256, eps=0.0)
          (relu): ReLU(inplace=True)
        )
      )
      (layer2): Sequential(
        (0): Bottleneck(
          (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(128, eps=0.0)
          (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(128, eps=0.0)
          (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(512, eps=0.0)
          (relu): ReLU(inplace=True)
          (downsample): Sequential(
            (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)
            (1): FrozenBatchNorm2d(512, eps=0.0)
          )
        )
        (1): Bottleneck(
          (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(128, eps=0.0)
          (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(128, eps=0.0)
          (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(512, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (2): Bottleneck(
          (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(128, eps=0.0)
          (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(128, eps=0.0)
          (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(512, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (3): Bottleneck(
          (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(128, eps=0.0)
          (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(128, eps=0.0)
          (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(512, eps=0.0)
          (relu): ReLU(inplace=True)
        )
      )
      (layer3): Sequential(
        (0): Bottleneck(
          (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
          (downsample): Sequential(
            (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False)
            (1): FrozenBatchNorm2d(1024, eps=0.0)
          )
        )
        (1): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (2): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (3): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (4): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (5): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
        )
      )
      (layer4): Sequential(
        (0): Bottleneck(
          (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(512, eps=0.0)
          (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(512, eps=0.0)
          (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(2048, eps=0.0)
          (relu): ReLU(inplace=True)
          (downsample): Sequential(
            (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False)
            (1): FrozenBatchNorm2d(2048, eps=0.0)
          )
        )
        (1): Bottleneck(
          (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(512, eps=0.0)
          (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(512, eps=0.0)
          (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(2048, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (2): Bottleneck(
          (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(512, eps=0.0)
          (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(512, eps=0.0)
          (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(2048, eps=0.0)
          (relu): ReLU(inplace=True)
        )
      )
    )
    (fpn): FeaturePyramidNetwork(
      (inner_blocks): ModuleList(
        (0): Conv2dNormActivation(
          (0): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1))
        )
        (1): Conv2dNormActivation(
          (0): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1))
        )
        (2): Conv2dNormActivation(
          (0): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1))
        )
        (3): Conv2dNormActivation(
          (0): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1))
        )
      )
      (layer_blocks): ModuleList(
        (0-3): 4 x Conv2dNormActivation(
          (0): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
        )
      )
      (extra_blocks): LastLevelMaxPool()
    )
  )
  (rpn): RegionProposalNetwork(
    (anchor_generator): AnchorGenerator()
    (head): RPNHead(
      (conv): Sequential(
        (0): Conv2dNormActivation(
          (0): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
          (1): ReLU(inplace=True)
        )
      )
      (cls_logits): Conv2d(256, 3, kernel_size=(1, 1), stride=(1, 1))
      (bbox_pred): Conv2d(256, 12, kernel_size=(1, 1), stride=(1, 1))
    )
  )
  (roi_heads): RoIHeads(
    (box_roi_pool): MultiScaleRoIAlign(featmap_names=['0', '1', '2', '3'], output_size=(7, 7), sampling_ratio=2)
    (box_head): TwoMLPHead(
      (fc6): Linear(in_features=12544, out_features=1024, bias=True)
      (fc7): Linear(in_features=1024, out_features=1024, bias=True)
    )
    (box_predictor): FastRCNNPredictor(
      (cls_score): Linear(in_features=1024, out_features=11, bias=True)
      (bbox_pred): Linear(in_features=1024, out_features=44, bias=True)
    )
  )
)

#rms prob
# to train on GPU if selected
device = torch.device('cuda') if torch.cuda.is_available() else torch.device('cpu')

# number of classes
num_classes = 11

# get the model using our helper function
model = get_object_detection_model(num_classes)

# move model to the right device
model.to(device)

# construct an optimizer
params = [p for p in model.parameters() if p.requires_grad]
optimizer = torch.optim.RMSprop(params, lr=0.001, weight_decay=0.0005)

# and a learning rate scheduler which decreases the learning rate by
# 10x every 3 epochs
lr_scheduler = torch.optim.lr_scheduler.StepLR(optimizer,
                                               step_size=3,
                                               gamma=0.1)

# training for 8 epochs # rmsprob
num_epochs = 15

for epoch in range(num_epochs):
    # training for one epoch
    train_one_epoch(model, optimizer, data_loader, device, epoch, print_freq=10)
    # update the learning rate
    lr_scheduler.step()
    # evaluate on the test dataset
    evaluate(model, data_loader_test, device=device)

Epoch: [0]  [  0/104]  eta: 0:04:00  lr: 0.000011  loss: 3.2039 (3.2039)  loss_classifier: 2.6346 (2.6346)  loss_box_reg: 0.2353 (0.2353)  loss_objectness: 0.3166 (0.3166)  loss_rpn_box_reg: 0.0173 (0.0173)  time: 2.3164  data: 0.8751  max mem: 6508
Epoch: [0]  [ 10/104]  eta: 0:01:58  lr: 0.000108  loss: 1.2988 (2.1030)  loss_classifier: 0.7136 (1.1584)  loss_box_reg: 0.2786 (0.2337)  loss_objectness: 0.3166 (0.6631)  loss_rpn_box_reg: 0.0292 (0.0478)  time: 1.2616  data: 0.1041  max mem: 6508
Epoch: [0]  [ 20/104]  eta: 0:01:44  lr: 0.000205  loss: 1.2517 (1.9053)  loss_classifier: 0.5494 (1.1040)  loss_box_reg: 0.2690 (0.2812)  loss_objectness: 0.2239 (0.4791)  loss_rpn_box_reg: 0.0292 (0.0410)  time: 1.1858  data: 0.0383  max mem: 6508
Epoch: [0]  [ 30/104]  eta: 0:01:27  lr: 0.000302  loss: 1.0549 (1.6985)  loss_classifier: 0.6310 (0.9926)  loss_box_reg: 0.2180 (0.2768)  loss_objectness: 0.1345 (0.3931)  loss_rpn_box_reg: 0.0252 (0.0360)  time: 1.1374  data: 0.0344  max mem: 6508
Epoch: [0]  [ 40/104]  eta: 0:01:13  lr: 0.000399  loss: 1.3226 (3.2637)  loss_classifier: 0.7431 (1.0957)  loss_box_reg: 0.2201 (0.4396)  loss_objectness: 0.2049 (1.5802)  loss_rpn_box_reg: 0.0295 (0.1482)  time: 1.0445  data: 0.0196  max mem: 6508
Epoch: [0]  [ 50/104]  eta: 0:01:00  lr: 0.000496  loss: 2.6148 (2047.7227)  loss_classifier: 1.1095 (320.6169)  loss_box_reg: 0.2412 (1473.0099)  loss_objectness: 0.6097 (235.9155)  loss_rpn_box_reg: 0.0607 (18.1803)  time: 1.0051  data: 0.0208  max mem: 6508
Epoch: [0]  [ 60/104]  eta: 0:00:48  lr: 0.000593  loss: 2.2350 (1712.3068)  loss_classifier: 0.6001 (268.1364)  loss_box_reg: 0.1415 (1231.5624)  loss_objectness: 0.6692 (197.3820)  loss_rpn_box_reg: 0.1377 (15.2260)  time: 0.9892  data: 0.0232  max mem: 6508
Epoch: [0]  [ 70/104]  eta: 0:00:36  lr: 0.000690  loss: 0.9199 (1471.2591)  loss_classifier: 0.3719 (230.4266)  loss_box_reg: 0.1484 (1058.1248)  loss_objectness: 0.3157 (169.6188)  loss_rpn_box_reg: 0.0653 (13.0889)  time: 1.0034  data: 0.0225  max mem: 6508
Epoch: [0]  [ 80/104]  eta: 0:00:25  lr: 0.000787  loss: 0.6987 (1289.6876)  loss_classifier: 0.2646 (202.0060)  loss_box_reg: 0.1104 (927.5039)  loss_objectness: 0.1908 (148.7011)  loss_rpn_box_reg: 0.0342 (11.4766)  time: 1.0148  data: 0.0234  max mem: 6508
Epoch: [0]  [ 90/104]  eta: 0:00:14  lr: 0.000884  loss: 0.5503 (1148.0251)  loss_classifier: 0.2011 (179.8337)  loss_box_reg: 0.1104 (825.5942)  loss_objectness: 0.1656 (132.3790)  loss_rpn_box_reg: 0.0246 (10.2181)  time: 1.0232  data: 0.0248  max mem: 6508
Epoch: [0]  [100/104]  eta: 0:00:04  lr: 0.000981  loss: 0.6267 (1034.4380)  loss_classifier: 0.2911 (162.0643)  loss_box_reg: 0.1404 (743.8696)  loss_objectness: 0.2006 (119.2944)  loss_rpn_box_reg: 0.0273 (9.2096)  time: 1.0243  data: 0.0203  max mem: 6508
Epoch: [0]  [103/104]  eta: 0:00:01  lr: 0.001000  loss: 0.6827 (1004.6290)  loss_classifier: 0.3058 (157.3996)  loss_box_reg: 0.1531 (722.4157)  loss_objectness: 0.2081 (115.8637)  loss_rpn_box_reg: 0.0323 (8.9500)  time: 1.0285  data: 0.0204  max mem: 6508
Epoch: [0] Total time: 0:01:50 (1.0649 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:33  model_time: 0.4742 (0.4742)  evaluator_time: 0.0052 (0.0052)  time: 1.2894  data: 0.7832  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4315 (0.4316)  evaluator_time: 0.0025 (0.0029)  time: 0.4587  data: 0.0188  max mem: 6508
Test: Total time: 0:00:12 (0.4944 s / it)
Averaged stats: model_time: 0.4315 (0.4316)  evaluator_time: 0.0025 (0.0029)
Accumulating evaluation results...
DONE (t=0.07s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
Epoch: [1]  [  0/104]  eta: 0:03:33  lr: 0.001000  loss: 1.4520 (1.4520)  loss_classifier: 0.6604 (0.6604)  loss_box_reg: 0.4759 (0.4759)  loss_objectness: 0.2343 (0.2343)  loss_rpn_box_reg: 0.0815 (0.0815)  time: 2.0520  data: 1.0032  max mem: 6508
Epoch: [1]  [ 10/104]  eta: 0:01:37  lr: 0.001000  loss: 258.9862 (150766802.6828)  loss_classifier: 127.6931 (34160226.4197)  loss_box_reg: 100.0327 (111223218.1034)  loss_objectness: 24.7305 (4924468.5647)  loss_rpn_box_reg: 8.8204 (458898.8178)  time: 1.0388  data: 0.1060  max mem: 6508
Epoch: [1]  [ 20/104]  eta: 0:01:21  lr: 0.001000  loss: 5199.8223 (110306276.1123)  loss_classifier: 2014.8184 (25663462.8065)  loss_box_reg: 2918.8621 (80794828.5285)  loss_objectness: 210.1121 (3472368.6373)  loss_rpn_box_reg: 56.0295 (375620.2806)  time: 0.9218  data: 0.0179  max mem: 6508
Epoch: [1]  [ 30/104]  eta: 0:01:10  lr: 0.001000  loss: 2652.1675 (74724978.8528)  loss_classifier: 1474.1609 (17385822.3312)  loss_box_reg: 1091.9497 (54732404.6968)  loss_objectness: 139.8561 (2352287.9434)  loss_rpn_box_reg: 30.5260 (254466.6862)  time: 0.9072  data: 0.0208  max mem: 6508
Epoch: [1]  [ 40/104]  eta: 0:01:00  lr: 0.001000  loss: 1029.4604 (56626685.1848)  loss_classifier: 611.0850 (13175208.6310)  loss_box_reg: 374.0753 (41479597.3672)  loss_objectness: 93.0817 (1779160.0837)  loss_rpn_box_reg: 24.7837 (192721.2135)  time: 0.8983  data: 0.0207  max mem: 6508
Epoch: [1]  [ 50/104]  eta: 0:00:50  lr: 0.001000  loss: 660.6842 (46107377.0981)  loss_classifier: 430.4982 (10690059.8973)  loss_box_reg: 89.7603 (33828148.1769)  loss_objectness: 88.0877 (1432500.1503)  loss_rpn_box_reg: 20.9473 (156670.5400)  time: 0.9033  data: 0.0195  max mem: 6508
Epoch: [1]  [ 60/104]  eta: 0:00:40  lr: 0.001000  loss: 426.3805 (38549403.6476)  loss_classifier: 143.6905 (8937839.6519)  loss_box_reg: 44.3117 (28282871.8341)  loss_objectness: 109.9760 (1197698.4676)  loss_rpn_box_reg: 24.6669 (130995.0873)  time: 0.9105  data: 0.0201  max mem: 6508
Epoch: [1]  [ 70/104]  eta: 0:00:31  lr: 0.001000  loss: 1013.4094 (33122822.6003)  loss_classifier: 778.9881 (7679878.7422)  loss_box_reg: 142.9680 (24301273.3163)  loss_objectness: 96.8082 (1029107.5885)  loss_rpn_box_reg: 24.6669 (112564.1501)  time: 0.8864  data: 0.0201  max mem: 6508
Epoch: [1]  [ 80/104]  eta: 0:00:22  lr: 0.001000  loss: 840.2881 (29035017.0729)  loss_classifier: 403.8030 (6732312.8815)  loss_box_reg: 67.4619 (21301945.7503)  loss_objectness: 109.2580 (902086.1737)  loss_rpn_box_reg: 32.2513 (98673.3166)  time: 0.8857  data: 0.0209  max mem: 6508
Epoch: [1]  [ 90/104]  eta: 0:00:12  lr: 0.001000  loss: 199.0619 (25849412.5666)  loss_classifier: 57.0748 (5994171.1934)  loss_box_reg: 25.5540 (18964375.8005)  loss_objectness: 109.2580 (803029.8781)  loss_rpn_box_reg: 20.1335 (87836.6283)  time: 0.9096  data: 0.0213  max mem: 6508
Epoch: [1]  [100/104]  eta: 0:00:03  lr: 0.001000  loss: 2028.7905 (23291665.0164)  loss_classifier: 410.3585 (5401432.9579)  loss_box_reg: 74.8636 (17087452.5200)  loss_objectness: 100.7363 (723622.5691)  loss_rpn_box_reg: 39.7727 (79157.8108)  time: 0.9073  data: 0.0197  max mem: 6508
Epoch: [1]  [103/104]  eta: 0:00:00  lr: 0.001000  loss: 2464.8901 (22620711.5484)  loss_classifier: 1564.9734 (5245974.8526)  loss_box_reg: 123.4845 (16595103.2782)  loss_objectness: 123.8959 (702756.3317)  loss_rpn_box_reg: 52.1484 (76877.9029)  time: 0.9007  data: 0.0198  max mem: 6508
Epoch: [1] Total time: 0:01:35 (0.9172 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:46  model_time: 0.5445 (0.5445)  evaluator_time: 0.0575 (0.0575)  time: 1.7704  data: 1.1407  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.3761 (0.3835)  evaluator_time: 0.0077 (0.0256)  time: 0.4286  data: 0.0189  max mem: 6508
Test: Total time: 0:00:12 (0.4859 s / it)
Averaged stats: model_time: 0.3761 (0.3835)  evaluator_time: 0.0077 (0.0256)
Accumulating evaluation results...
DONE (t=0.21s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
Epoch: [2]  [  0/104]  eta: 0:04:04  lr: 0.001000  loss: 3665.4360 (3665.4360)  loss_classifier: 3275.9192 (3275.9192)  loss_box_reg: 63.6991 (63.6991)  loss_objectness: 234.6715 (234.6715)  loss_rpn_box_reg: 91.1466 (91.1466)  time: 2.3475  data: 1.2587  max mem: 6508
Epoch: [2]  [ 10/104]  eta: 0:01:38  lr: 0.001000  loss: 1106.7623 (1422.6658)  loss_classifier: 562.6231 (1053.5947)  loss_box_reg: 90.5225 (282.9638)  loss_objectness: 47.2851 (66.0814)  loss_rpn_box_reg: 10.7598 (20.0259)  time: 1.0521  data: 0.1272  max mem: 6508
Epoch: [2]  [ 20/104]  eta: 0:01:23  lr: 0.001000  loss: 472.3109 (1020.1505)  loss_classifier: 244.4862 (661.8328)  loss_box_reg: 85.9072 (273.9741)  loss_objectness: 43.4546 (62.3635)  loss_rpn_box_reg: 14.1286 (21.9800)  time: 0.9309  data: 0.0193  max mem: 6508
Epoch: [2]  [ 30/104]  eta: 0:01:12  lr: 0.001000  loss: 121.1059 (731.7680)  loss_classifier: 41.7808 (463.9359)  loss_box_reg: 22.7789 (198.2694)  loss_objectness: 30.6498 (50.8209)  loss_rpn_box_reg: 12.7459 (18.7418)  time: 0.9380  data: 0.0235  max mem: 6508
Epoch: [2]  [ 40/104]  eta: 0:01:01  lr: 0.001000  loss: 133.5830 (634.6437)  loss_classifier: 69.0693 (380.3669)  loss_box_reg: 33.9993 (188.5627)  loss_objectness: 25.5532 (47.8302)  loss_rpn_box_reg: 10.1752 (17.8839)  time: 0.9087  data: 0.0206  max mem: 6508
Epoch: [2]  [ 50/104]  eta: 0:00:50  lr: 0.001000  loss: 273.1735 (615.1750)  loss_classifier: 69.0693 (345.7360)  loss_box_reg: 79.6419 (199.5584)  loss_objectness: 35.9543 (50.6717)  loss_rpn_box_reg: 12.6669 (19.2088)  time: 0.8914  data: 0.0195  max mem: 6508
Epoch: [2]  [ 60/104]  eta: 0:00:41  lr: 0.001000  loss: 365.2442 (3485.2267)  loss_classifier: 124.5114 (1472.7287)  loss_box_reg: 80.6085 (1905.5887)  loss_objectness: 61.5172 (79.8159)  loss_rpn_box_reg: 25.2533 (27.0934)  time: 0.8913  data: 0.0218  max mem: 6508
Epoch: [2]  [ 70/104]  eta: 0:00:31  lr: 0.001000  loss: 594.5722 (3400.5566)  loss_classifier: 283.9956 (1487.2163)  loss_box_reg: 191.6195 (1795.6577)  loss_objectness: 125.6146 (88.4065)  loss_rpn_box_reg: 37.0607 (29.2761)  time: 0.8710  data: 0.0222  max mem: 6508
Epoch: [2]  [ 80/104]  eta: 0:00:22  lr: 0.001000  loss: 490.2880 (3256.1005)  loss_classifier: 126.4839 (1455.8884)  loss_box_reg: 82.9989 (1676.2213)  loss_objectness: 125.6146 (93.2964)  loss_rpn_box_reg: 35.4058 (30.6944)  time: 0.8967  data: 0.0251  max mem: 6508
Epoch: [2]  [ 90/104]  eta: 0:00:12  lr: 0.001000  loss: 517.9962 (3362.4710)  loss_classifier: 175.8158 (1395.0213)  loss_box_reg: 82.9989 (1828.0077)  loss_objectness: 131.6742 (106.7316)  loss_rpn_box_reg: 31.9363 (32.7103)  time: 0.9032  data: 0.0249  max mem: 6508
Epoch: [2]  [100/104]  eta: 0:00:03  lr: 0.001000  loss: 480.0167 (3061.3139)  loss_classifier: 99.8471 (1265.5824)  loss_box_reg: 123.9330 (1659.4183)  loss_objectness: 77.2580 (104.0533)  loss_rpn_box_reg: 23.2427 (32.2599)  time: 0.8693  data: 0.0208  max mem: 6508
Epoch: [2]  [103/104]  eta: 0:00:00  lr: 0.001000  loss: 315.5242 (2983.3835)  loss_classifier: 89.9003 (1231.4645)  loss_box_reg: 120.3518 (1615.6943)  loss_objectness: 61.8435 (104.0673)  loss_rpn_box_reg: 14.4586 (32.1574)  time: 0.8691  data: 0.0202  max mem: 6508
Epoch: [2] Total time: 0:01:35 (0.9136 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:31  model_time: 0.4697 (0.4697)  evaluator_time: 0.0529 (0.0529)  time: 1.2306  data: 0.6546  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.3741 (0.3791)  evaluator_time: 0.0056 (0.0077)  time: 0.4157  data: 0.0241  max mem: 6508
Test: Total time: 0:00:11 (0.4538 s / it)
Averaged stats: model_time: 0.3741 (0.3791)  evaluator_time: 0.0056 (0.0077)
Accumulating evaluation results...
DONE (t=0.09s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
Epoch: [3]  [  0/104]  eta: 0:03:05  lr: 0.000100  loss: 304.2431 (304.2431)  loss_classifier: 84.9447 (84.9447)  loss_box_reg: 82.5549 (82.5549)  loss_objectness: 105.4210 (105.4210)  loss_rpn_box_reg: 31.3225 (31.3225)  time: 1.7828  data: 0.9015  max mem: 6508
Epoch: [3]  [ 10/104]  eta: 0:01:30  lr: 0.000100  loss: 272.1476 (248.5491)  loss_classifier: 39.4093 (47.7809)  loss_box_reg: 107.1224 (101.2873)  loss_objectness: 75.0196 (71.9039)  loss_rpn_box_reg: 12.8808 (27.5771)  time: 0.9625  data: 0.0974  max mem: 6508
Epoch: [3]  [ 20/104]  eta: 0:01:17  lr: 0.000100  loss: 265.4117 (264.0045)  loss_classifier: 43.5565 (56.6697)  loss_box_reg: 88.7669 (94.0715)  loss_objectness: 75.0196 (82.9571)  loss_rpn_box_reg: 13.7246 (30.3063)  time: 0.8823  data: 0.0180  max mem: 6508
Epoch: [3]  [ 30/104]  eta: 0:01:07  lr: 0.000100  loss: 169.6590 (232.1553)  loss_classifier: 35.6502 (46.1811)  loss_box_reg: 55.9581 (74.7482)  loss_objectness: 66.9722 (80.4461)  loss_rpn_box_reg: 25.9141 (30.7800)  time: 0.8803  data: 0.0193  max mem: 6508
Epoch: [3]  [ 40/104]  eta: 0:00:57  lr: 0.000100  loss: 148.1319 (219.2072)  loss_classifier: 14.4577 (40.4276)  loss_box_reg: 39.1762 (69.0179)  loss_objectness: 67.2271 (79.7802)  loss_rpn_box_reg: 28.7024 (29.9815)  time: 0.8741  data: 0.0213  max mem: 6508
Epoch: [3]  [ 50/104]  eta: 0:00:48  lr: 0.000100  loss: 142.2928 (205.5259)  loss_classifier: 12.6049 (34.9610)  loss_box_reg: 47.2105 (64.0626)  loss_objectness: 55.7569 (76.3784)  loss_rpn_box_reg: 22.1210 (30.1239)  time: 0.8627  data: 0.0208  max mem: 6508
Epoch: [3]  [ 60/104]  eta: 0:00:38  lr: 0.000100  loss: 133.3797 (195.1930)  loss_classifier: 10.6868 (31.3746)  loss_box_reg: 38.1073 (60.6654)  loss_objectness: 46.0740 (73.7778)  loss_rpn_box_reg: 17.6979 (29.3751)  time: 0.8546  data: 0.0194  max mem: 6508
Epoch: [3]  [ 70/104]  eta: 0:00:29  lr: 0.000100  loss: 141.7962 (192.2237)  loss_classifier: 8.9238 (28.4367)  loss_box_reg: 38.1073 (59.2055)  loss_objectness: 41.5079 (74.2889)  loss_rpn_box_reg: 19.9482 (30.2926)  time: 0.8553  data: 0.0202  max mem: 6508
Epoch: [3]  [ 80/104]  eta: 0:00:21  lr: 0.000100  loss: 97.7690 (183.0945)  loss_classifier: 9.0629 (26.6522)  loss_box_reg: 35.0960 (56.4913)  loss_objectness: 40.6501 (70.6300)  loss_rpn_box_reg: 13.6175 (29.3210)  time: 0.8527  data: 0.0197  max mem: 6508
Epoch: [3]  [ 90/104]  eta: 0:00:12  lr: 0.000100  loss: 89.6644 (177.9917)  loss_classifier: 10.7990 (25.1199)  loss_box_reg: 29.3037 (56.3916)  loss_objectness: 32.6587 (68.0665)  loss_rpn_box_reg: 12.3882 (28.4137)  time: 0.8554  data: 0.0211  max mem: 6508
Epoch: [3]  [100/104]  eta: 0:00:03  lr: 0.000100  loss: 91.6710 (171.5448)  loss_classifier: 10.7990 (23.8506)  loss_box_reg: 29.0876 (54.6277)  loss_objectness: 41.3260 (65.3209)  loss_rpn_box_reg: 13.0908 (27.7456)  time: 0.8577  data: 0.0206  max mem: 6508
Epoch: [3]  [103/104]  eta: 0:00:00  lr: 0.000100  loss: 91.6710 (169.4891)  loss_classifier: 10.7990 (23.4692)  loss_box_reg: 25.1067 (54.1283)  loss_objectness: 35.2736 (64.5671)  loss_rpn_box_reg: 13.0908 (27.3245)  time: 0.8590  data: 0.0207  max mem: 6508
Epoch: [3] Total time: 0:01:30 (0.8744 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:37  model_time: 0.4409 (0.4409)  evaluator_time: 0.0071 (0.0071)  time: 1.4601  data: 0.9932  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.3669 (0.3692)  evaluator_time: 0.0024 (0.0029)  time: 0.3942  data: 0.0186  max mem: 6508
Test: Total time: 0:00:11 (0.4410 s / it)
Averaged stats: model_time: 0.3669 (0.3692)  evaluator_time: 0.0024 (0.0029)
Accumulating evaluation results...
DONE (t=0.07s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
Epoch: [4]  [  0/104]  eta: 0:04:17  lr: 0.000100  loss: 66.1785 (66.1785)  loss_classifier: 2.6410 (2.6410)  loss_box_reg: 28.5581 (28.5581)  loss_objectness: 26.6346 (26.6346)  loss_rpn_box_reg: 8.3447 (8.3447)  time: 2.4786  data: 1.2914  max mem: 6508
Epoch: [4]  [ 10/104]  eta: 0:01:36  lr: 0.000100  loss: 110.0946 (153.1176)  loss_classifier: 9.2976 (12.9349)  loss_box_reg: 35.5208 (49.3572)  loss_objectness: 35.9742 (61.9151)  loss_rpn_box_reg: 16.9071 (28.9103)  time: 1.0231  data: 0.1354  max mem: 6508
Epoch: [4]  [ 20/104]  eta: 0:01:20  lr: 0.000100  loss: 110.0946 (135.4778)  loss_classifier: 9.2976 (12.4413)  loss_box_reg: 35.5208 (45.8977)  loss_objectness: 36.0392 (52.7461)  loss_rpn_box_reg: 13.6858 (24.3927)  time: 0.8864  data: 0.0213  max mem: 6508
Epoch: [4]  [ 30/104]  eta: 0:01:09  lr: 0.000100  loss: 109.8829 (125.2450)  loss_classifier: 11.8280 (14.1573)  loss_box_reg: 24.4844 (41.4060)  loss_objectness: 40.7744 (47.4909)  loss_rpn_box_reg: 10.3591 (22.1907)  time: 0.8879  data: 0.0227  max mem: 6508
Epoch: [4]  [ 40/104]  eta: 0:00:58  lr: 0.000100  loss: 83.0998 (115.2337)  loss_classifier: 12.7831 (14.0151)  loss_box_reg: 21.4768 (36.6713)  loss_objectness: 29.0046 (44.1066)  loss_rpn_box_reg: 10.4131 (20.4406)  time: 0.8762  data: 0.0227  max mem: 6508
Epoch: [4]  [ 50/104]  eta: 0:00:49  lr: 0.000100  loss: 79.7844 (112.6534)  loss_classifier: 12.8533 (14.6672)  loss_box_reg: 15.9272 (32.9115)  loss_objectness: 31.7846 (43.8854)  loss_rpn_box_reg: 12.3448 (21.1893)  time: 0.8691  data: 0.0226  max mem: 6508
Epoch: [4]  [ 60/104]  eta: 0:00:39  lr: 0.000100  loss: 88.3089 (110.2105)  loss_classifier: 14.3843 (14.4013)  loss_box_reg: 14.1213 (30.6656)  loss_objectness: 37.6492 (44.1127)  loss_rpn_box_reg: 14.8587 (21.0310)  time: 0.8636  data: 0.0221  max mem: 6508
Epoch: [4]  [ 70/104]  eta: 0:00:30  lr: 0.000100  loss: 76.0205 (104.6302)  loss_classifier: 5.5285 (12.7650)  loss_box_reg: 4.2127 (26.5244)  loss_objectness: 37.6492 (43.9946)  loss_rpn_box_reg: 16.4864 (21.3462)  time: 0.8680  data: 0.0243  max mem: 6508
Epoch: [4]  [ 80/104]  eta: 0:00:21  lr: 0.000100  loss: 60.9173 (99.6835)  loss_classifier: 0.5171 (11.2307)  loss_box_reg: 0.5805 (23.3846)  loss_objectness: 37.0190 (44.0343)  loss_rpn_box_reg: 17.7755 (21.0340)  time: 0.8956  data: 0.0250  max mem: 6508
Epoch: [4]  [ 90/104]  eta: 0:00:12  lr: 0.000100  loss: 47.0268 (95.8740)  loss_classifier: 0.2698 (10.0324)  loss_box_reg: 0.3479 (20.8609)  loss_objectness: 36.3957 (43.7518)  loss_rpn_box_reg: 11.3258 (21.2289)  time: 0.9245  data: 0.0208  max mem: 6508
Epoch: [4]  [100/104]  eta: 0:00:03  lr: 0.000100  loss: 46.4205 (92.1205)  loss_classifier: 0.2538 (9.0769)  loss_box_reg: 0.3883 (18.8635)  loss_objectness: 34.8001 (43.2130)  loss_rpn_box_reg: 10.7662 (20.9672)  time: 0.9419  data: 0.0185  max mem: 6508
Epoch: [4]  [103/104]  eta: 0:00:00  lr: 0.000100  loss: 46.4205 (91.3613)  loss_classifier: 0.2067 (8.8189)  loss_box_reg: 0.3031 (18.3242)  loss_objectness: 34.8001 (43.2987)  loss_rpn_box_reg: 11.2495 (20.9195)  time: 0.9473  data: 0.0181  max mem: 6508
Epoch: [4] Total time: 0:01:34 (0.9111 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:31  model_time: 0.4938 (0.4938)  evaluator_time: 0.0210 (0.0210)  time: 1.2012  data: 0.6711  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.3945 (0.3990)  evaluator_time: 0.0087 (0.0117)  time: 0.4304  data: 0.0188  max mem: 6508
Test: Total time: 0:00:12 (0.4672 s / it)
Averaged stats: model_time: 0.3945 (0.3990)  evaluator_time: 0.0087 (0.0117)
Accumulating evaluation results...
DONE (t=0.10s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
Epoch: [5]  [  0/104]  eta: 0:03:08  lr: 0.000100  loss: 20.0014 (20.0014)  loss_classifier: 0.1053 (0.1053)  loss_box_reg: 0.0933 (0.0933)  loss_objectness: 17.1295 (17.1295)  loss_rpn_box_reg: 2.6733 (2.6733)  time: 1.8093  data: 0.7300  max mem: 6508
Epoch: [5]  [ 10/104]  eta: 0:01:41  lr: 0.000100  loss: 41.9883 (67.5856)  loss_classifier: 0.2323 (0.4990)  loss_box_reg: 0.5077 (0.8016)  loss_objectness: 31.6889 (44.3222)  loss_rpn_box_reg: 8.5974 (21.9628)  time: 1.0764  data: 0.0864  max mem: 6508
Epoch: [5]  [ 20/104]  eta: 0:01:27  lr: 0.000100  loss: 41.9883 (74.8238)  loss_classifier: 0.2920 (0.4492)  loss_box_reg: 0.6207 (0.7669)  loss_objectness: 31.6889 (48.9295)  loss_rpn_box_reg: 9.0488 (24.6782)  time: 1.0055  data: 0.0221  max mem: 6508
Epoch: [5]  [ 30/104]  eta: 0:01:15  lr: 0.000100  loss: 36.4438 (63.3653)  loss_classifier: 0.2598 (0.4511)  loss_box_reg: 0.5993 (0.7657)  loss_objectness: 26.6947 (41.4684)  loss_rpn_box_reg: 9.4150 (20.6801)  time: 0.9988  data: 0.0209  max mem: 6508
Epoch: [5]  [ 40/104]  eta: 0:01:04  lr: 0.000100  loss: 35.3091 (63.8564)  loss_classifier: 0.3117 (0.5346)  loss_box_reg: 0.8865 (0.9180)  loss_objectness: 24.7180 (41.4591)  loss_rpn_box_reg: 8.4455 (20.9447)  time: 0.9850  data: 0.0208  max mem: 6508
Epoch: [5]  [ 50/104]  eta: 0:00:54  lr: 0.000100  loss: 45.7666 (63.1234)  loss_classifier: 0.8353 (0.7369)  loss_box_reg: 0.7469 (0.8398)  loss_objectness: 30.9653 (40.4953)  loss_rpn_box_reg: 12.1707 (21.0514)  time: 0.9794  data: 0.0232  max mem: 6508
Epoch: [5]  [ 60/104]  eta: 0:00:43  lr: 0.000100  loss: 47.3924 (63.5978)  loss_classifier: 0.8437 (0.7450)  loss_box_reg: 0.3043 (0.7829)  loss_objectness: 32.3307 (41.0266)  loss_rpn_box_reg: 13.1207 (21.0433)  time: 0.9675  data: 0.0219  max mem: 6508
Epoch: [5]  [ 70/104]  eta: 0:00:33  lr: 0.000100  loss: 48.2683 (62.4787)  loss_classifier: 0.7768 (0.7573)  loss_box_reg: 0.5655 (0.8049)  loss_objectness: 32.3307 (40.3854)  loss_rpn_box_reg: 13.3730 (20.5311)  time: 0.9623  data: 0.0232  max mem: 6508
Epoch: [5]  [ 80/104]  eta: 0:00:23  lr: 0.000100  loss: 47.3047 (60.7057)  loss_classifier: 0.6391 (0.7672)  loss_box_reg: 0.7198 (0.7969)  loss_objectness: 31.1596 (39.3882)  loss_rpn_box_reg: 13.4515 (19.7535)  time: 0.9573  data: 0.0253  max mem: 6508
Epoch: [5]  [ 90/104]  eta: 0:00:13  lr: 0.000100  loss: 40.0958 (58.2975)  loss_classifier: 0.5869 (0.7933)  loss_box_reg: 0.6141 (0.8455)  loss_objectness: 28.8788 (38.0051)  loss_rpn_box_reg: 12.1153 (18.6536)  time: 0.9632  data: 0.0246  max mem: 6508
Epoch: [5]  [100/104]  eta: 0:00:03  lr: 0.000100  loss: 40.0958 (57.9070)  loss_classifier: 0.7374 (0.8084)  loss_box_reg: 0.6148 (0.8468)  loss_objectness: 26.4502 (37.8837)  loss_rpn_box_reg: 9.1422 (18.3681)  time: 0.9601  data: 0.0218  max mem: 6508
Epoch: [5]  [103/104]  eta: 0:00:00  lr: 0.000100  loss: 37.3095 (57.5049)  loss_classifier: 0.7374 (0.8197)  loss_box_reg: 0.6622 (0.8917)  loss_objectness: 26.3762 (37.6363)  loss_rpn_box_reg: 9.1422 (18.1572)  time: 0.9572  data: 0.0222  max mem: 6508
Epoch: [5] Total time: 0:01:42 (0.9836 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:32  model_time: 0.4740 (0.4740)  evaluator_time: 0.0147 (0.0147)  time: 1.2605  data: 0.7454  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.3786 (0.3810)  evaluator_time: 0.0057 (0.0071)  time: 0.4114  data: 0.0203  max mem: 6508
Test: Total time: 0:00:11 (0.4494 s / it)
Averaged stats: model_time: 0.3786 (0.3810)  evaluator_time: 0.0057 (0.0071)
Accumulating evaluation results...
DONE (t=0.12s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
Epoch: [6]  [  0/104]  eta: 0:03:57  lr: 0.000010  loss: 24.8814 (24.8814)  loss_classifier: 0.8751 (0.8751)  loss_box_reg: 1.3191 (1.3191)  loss_objectness: 18.5028 (18.5028)  loss_rpn_box_reg: 4.1844 (4.1844)  time: 2.2849  data: 1.1360  max mem: 6508
Epoch: [6]  [ 10/104]  eta: 0:01:41  lr: 0.000010  loss: 37.0075 (51.6789)  loss_classifier: 0.8751 (1.3198)  loss_box_reg: 1.0455 (1.6272)  loss_objectness: 23.6202 (29.5659)  loss_rpn_box_reg: 12.6925 (19.1660)  time: 1.0767  data: 0.1201  max mem: 6508
Epoch: [6]  [ 20/104]  eta: 0:01:25  lr: 0.000010  loss: 44.0930 (62.7234)  loss_classifier: 0.8575 (1.1503)  loss_box_reg: 0.8651 (1.4286)  loss_objectness: 27.5672 (38.6000)  loss_rpn_box_reg: 13.2039 (21.5444)  time: 0.9558  data: 0.0206  max mem: 6508
Epoch: [6]  [ 30/104]  eta: 0:01:13  lr: 0.000010  loss: 57.8508 (66.9983)  loss_classifier: 0.9510 (1.3345)  loss_box_reg: 0.9510 (1.5951)  loss_objectness: 35.4797 (41.0589)  loss_rpn_box_reg: 15.4171 (23.0098)  time: 0.9395  data: 0.0226  max mem: 6508
Epoch: [6]  [ 40/104]  eta: 0:01:01  lr: 0.000010  loss: 36.9282 (61.7737)  loss_classifier: 1.2070 (1.3313)  loss_box_reg: 0.8505 (1.5105)  loss_objectness: 25.2796 (38.4129)  loss_rpn_box_reg: 10.4051 (20.5190)  time: 0.9082  data: 0.0216  max mem: 6508
Epoch: [6]  [ 50/104]  eta: 0:00:51  lr: 0.000010  loss: 32.7680 (58.1379)  loss_classifier: 1.2070 (1.3677)  loss_box_reg: 1.2339 (1.6008)  loss_objectness: 21.0982 (36.3104)  loss_rpn_box_reg: 5.5364 (18.8590)  time: 0.8981  data: 0.0225  max mem: 6508
Epoch: [6]  [ 60/104]  eta: 0:00:41  lr: 0.000010  loss: 47.7588 (57.6452)  loss_classifier: 1.3659 (1.3310)  loss_box_reg: 1.7621 (1.6372)  loss_objectness: 30.3292 (36.0883)  loss_rpn_box_reg: 9.6681 (18.5887)  time: 0.9035  data: 0.0237  max mem: 6508
Epoch: [6]  [ 70/104]  eta: 0:00:31  lr: 0.000010  loss: 51.2820 (58.2520)  loss_classifier: 1.3659 (1.4285)  loss_box_reg: 1.9540 (1.8565)  loss_objectness: 30.8757 (36.1639)  loss_rpn_box_reg: 14.9676 (18.8032)  time: 0.8961  data: 0.0212  max mem: 6508
Epoch: [6]  [ 80/104]  eta: 0:00:22  lr: 0.000010  loss: 35.5225 (57.1742)  loss_classifier: 1.2666 (1.4121)  loss_box_reg: 1.8910 (1.8212)  loss_objectness: 23.9973 (35.5701)  loss_rpn_box_reg: 8.0572 (18.3709)  time: 0.8954  data: 0.0219  max mem: 6508
Epoch: [6]  [ 90/104]  eta: 0:00:13  lr: 0.000010  loss: 50.4645 (59.3462)  loss_classifier: 1.0201 (1.4707)  loss_box_reg: 0.8960 (1.8823)  loss_objectness: 31.3237 (37.0268)  loss_rpn_box_reg: 11.1851 (18.9664)  time: 0.9084  data: 0.0241  max mem: 6508
Epoch: [6]  [100/104]  eta: 0:00:03  lr: 0.000010  loss: 34.5830 (56.6741)  loss_classifier: 1.1253 (1.4483)  loss_box_reg: 1.0179 (1.9148)  loss_objectness: 25.9233 (35.5759)  loss_rpn_box_reg: 8.6166 (17.7352)  time: 0.9026  data: 0.0213  max mem: 6508
Epoch: [6]  [103/104]  eta: 0:00:00  lr: 0.000010  loss: 34.5830 (56.2563)  loss_classifier: 1.0971 (1.4783)  loss_box_reg: 1.0179 (1.9213)  loss_objectness: 25.9233 (35.3313)  loss_rpn_box_reg: 7.3149 (17.5255)  time: 0.9046  data: 0.0213  max mem: 6508
Epoch: [6] Total time: 0:01:36 (0.9286 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:45  model_time: 0.5394 (0.5394)  evaluator_time: 0.0150 (0.0150)  time: 1.7539  data: 1.1803  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.3717 (0.3802)  evaluator_time: 0.0048 (0.0061)  time: 0.4023  data: 0.0195  max mem: 6508
Test: Total time: 0:00:12 (0.4624 s / it)
Averaged stats: model_time: 0.3717 (0.3802)  evaluator_time: 0.0048 (0.0061)
Accumulating evaluation results...
DONE (t=0.06s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
Epoch: [7]  [  0/104]  eta: 0:03:51  lr: 0.000010  loss: 87.8664 (87.8664)  loss_classifier: 0.7757 (0.7757)  loss_box_reg: 0.3214 (0.3214)  loss_objectness: 60.4727 (60.4727)  loss_rpn_box_reg: 26.2966 (26.2966)  time: 2.2264  data: 1.0773  max mem: 6508
Epoch: [7]  [ 10/104]  eta: 0:01:37  lr: 0.000010  loss: 47.3212 (64.1296)  loss_classifier: 1.4586 (2.0382)  loss_box_reg: 1.4097 (2.1041)  loss_objectness: 27.7213 (40.2605)  loss_rpn_box_reg: 14.9454 (19.7268)  time: 1.0400  data: 0.1201  max mem: 6508
Epoch: [7]  [ 20/104]  eta: 0:01:22  lr: 0.000010  loss: 39.5393 (58.7766)  loss_classifier: 1.2743 (2.0464)  loss_box_reg: 1.6440 (2.6563)  loss_objectness: 27.7213 (36.5018)  loss_rpn_box_reg: 8.0800 (17.5721)  time: 0.9220  data: 0.0242  max mem: 6508
Epoch: [7]  [ 30/104]  eta: 0:01:11  lr: 0.000010  loss: 38.5752 (53.8027)  loss_classifier: 0.9598 (1.7169)  loss_box_reg: 1.0913 (2.1371)  loss_objectness: 28.3370 (34.1707)  loss_rpn_box_reg: 8.0800 (15.7781)  time: 0.9190  data: 0.0219  max mem: 6508
Epoch: [7]  [ 40/104]  eta: 0:01:00  lr: 0.000010  loss: 42.6587 (56.0816)  loss_classifier: 0.9598 (1.6737)  loss_box_reg: 0.6641 (2.0950)  loss_objectness: 29.4402 (35.0810)  loss_rpn_box_reg: 11.0174 (17.2319)  time: 0.9084  data: 0.0201  max mem: 6508
Epoch: [7]  [ 50/104]  eta: 0:00:50  lr: 0.000010  loss: 50.4157 (58.5879)  loss_classifier: 1.0065 (1.7349)  loss_box_reg: 0.9827 (2.0586)  loss_objectness: 33.9286 (36.9820)  loss_rpn_box_reg: 15.4339 (17.8124)  time: 0.8982  data: 0.0208  max mem: 6508
Epoch: [7]  [ 60/104]  eta: 0:00:40  lr: 0.000010  loss: 34.5905 (55.2556)  loss_classifier: 0.7998 (1.6528)  loss_box_reg: 0.9682 (1.9844)  loss_objectness: 24.7950 (34.9693)  loss_rpn_box_reg: 7.9839 (16.6490)  time: 0.8985  data: 0.0214  max mem: 6508
Epoch: [7]  [ 70/104]  eta: 0:00:31  lr: 0.000010  loss: 29.1215 (55.7689)  loss_classifier: 0.7629 (1.6086)  loss_box_reg: 1.5312 (2.0877)  loss_objectness: 19.9744 (34.9326)  loss_rpn_box_reg: 5.5967 (17.1400)  time: 0.9043  data: 0.0213  max mem: 6508
Epoch: [7]  [ 80/104]  eta: 0:00:22  lr: 0.000010  loss: 35.2129 (55.1393)  loss_classifier: 0.9716 (1.5208)  loss_box_reg: 1.5312 (2.0082)  loss_objectness: 23.3236 (34.7006)  loss_rpn_box_reg: 6.5975 (16.9096)  time: 0.9068  data: 0.0211  max mem: 6508
Epoch: [7]  [ 90/104]  eta: 0:00:12  lr: 0.000010  loss: 37.3594 (55.3226)  loss_classifier: 0.9789 (1.5148)  loss_box_reg: 1.5428 (2.0304)  loss_objectness: 25.0347 (34.6903)  loss_rpn_box_reg: 10.9191 (17.0871)  time: 0.8987  data: 0.0203  max mem: 6508
Epoch: [7]  [100/104]  eta: 0:00:03  lr: 0.000010  loss: 38.2908 (55.4929)  loss_classifier: 1.1156 (1.5634)  loss_box_reg: 1.6116 (2.0739)  loss_objectness: 23.8722 (34.6085)  loss_rpn_box_reg: 8.4591 (17.2470)  time: 0.8915  data: 0.0194  max mem: 6508
Epoch: [7]  [103/104]  eta: 0:00:00  lr: 0.000010  loss: 48.2291 (56.1940)  loss_classifier: 1.1156 (1.5593)  loss_box_reg: 1.1863 (2.0916)  loss_objectness: 25.4885 (34.9871)  loss_rpn_box_reg: 11.4533 (17.5560)  time: 0.8940  data: 0.0189  max mem: 6508
Epoch: [7] Total time: 0:01:35 (0.9197 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:32  model_time: 0.4474 (0.4474)  evaluator_time: 0.0151 (0.0151)  time: 1.2321  data: 0.7581  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.3696 (0.3729)  evaluator_time: 0.0054 (0.0060)  time: 0.4167  data: 0.0345  max mem: 6508
Test: Total time: 0:00:11 (0.4516 s / it)
Averaged stats: model_time: 0.3696 (0.3729)  evaluator_time: 0.0054 (0.0060)
Accumulating evaluation results...
DONE (t=0.11s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
Epoch: [8]  [  0/104]  eta: 0:03:19  lr: 0.000010  loss: 15.9957 (15.9957)  loss_classifier: 0.3973 (0.3973)  loss_box_reg: 0.4962 (0.4962)  loss_objectness: 11.7689 (11.7689)  loss_rpn_box_reg: 3.3333 (3.3333)  time: 1.9187  data: 0.9615  max mem: 6508
Epoch: [8]  [ 10/104]  eta: 0:01:35  lr: 0.000010  loss: 32.1010 (56.6887)  loss_classifier: 1.5712 (1.5366)  loss_box_reg: 1.5819 (2.2572)  loss_objectness: 18.2520 (32.7409)  loss_rpn_box_reg: 7.2458 (20.1540)  time: 1.0139  data: 0.1060  max mem: 6508
Epoch: [8]  [ 20/104]  eta: 0:01:22  lr: 0.000010  loss: 32.2809 (49.5535)  loss_classifier: 0.6473 (1.2144)  loss_box_reg: 1.5546 (2.0191)  loss_objectness: 22.6618 (30.3870)  loss_rpn_box_reg: 6.8732 (15.9329)  time: 0.9337  data: 0.0210  max mem: 6508
Epoch: [8]  [ 30/104]  eta: 0:01:11  lr: 0.000010  loss: 33.1148 (50.5814)  loss_classifier: 0.6470 (1.1718)  loss_box_reg: 1.1354 (1.9984)  loss_objectness: 24.9701 (31.8476)  loss_rpn_box_reg: 5.7260 (15.5636)  time: 0.9408  data: 0.0207  max mem: 6508
Epoch: [8]  [ 40/104]  eta: 0:01:01  lr: 0.000010  loss: 38.0942 (49.5056)  loss_classifier: 0.7300 (1.1300)  loss_box_reg: 1.1214 (1.8952)  loss_objectness: 25.0311 (30.6048)  loss_rpn_box_reg: 8.7146 (15.8756)  time: 0.9374  data: 0.0220  max mem: 6508
Epoch: [8]  [ 50/104]  eta: 0:00:51  lr: 0.000010  loss: 36.8604 (51.6394)  loss_classifier: 0.9232 (1.1640)  loss_box_reg: 1.1698 (1.8768)  loss_objectness: 24.8990 (32.1092)  loss_rpn_box_reg: 8.7146 (16.4894)  time: 0.9236  data: 0.0228  max mem: 6508
Epoch: [8]  [ 60/104]  eta: 0:00:41  lr: 0.000010  loss: 44.3082 (56.5265)  loss_classifier: 0.9232 (1.1949)  loss_box_reg: 1.1370 (1.9806)  loss_objectness: 30.9550 (35.3398)  loss_rpn_box_reg: 9.3121 (18.0113)  time: 0.8972  data: 0.0205  max mem: 6508
Epoch: [8]  [ 70/104]  eta: 0:00:31  lr: 0.000010  loss: 56.3628 (57.0742)  loss_classifier: 1.0805 (1.2302)  loss_box_reg: 1.1370 (2.1024)  loss_objectness: 33.9098 (35.5870)  loss_rpn_box_reg: 16.4104 (18.1545)  time: 0.8856  data: 0.0213  max mem: 6508
Epoch: [8]  [ 80/104]  eta: 0:00:22  lr: 0.000010  loss: 39.0102 (54.4213)  loss_classifier: 1.2735 (1.3293)  loss_box_reg: 1.3677 (2.2665)  loss_objectness: 26.5310 (33.9587)  loss_rpn_box_reg: 7.9361 (16.8668)  time: 0.8897  data: 0.0229  max mem: 6508
Epoch: [8]  [ 90/104]  eta: 0:00:12  lr: 0.000010  loss: 34.5572 (54.0793)  loss_classifier: 0.9454 (1.2863)  loss_box_reg: 1.1661 (2.1681)  loss_objectness: 22.4779 (33.4381)  loss_rpn_box_reg: 8.1952 (17.1868)  time: 0.8948  data: 0.0215  max mem: 6508
Epoch: [8]  [100/104]  eta: 0:00:03  lr: 0.000010  loss: 49.5937 (55.8098)  loss_classifier: 0.6847 (1.3095)  loss_box_reg: 1.5164 (2.2174)  loss_objectness: 33.7459 (34.5916)  loss_rpn_box_reg: 12.6412 (17.6912)  time: 0.8954  data: 0.0203  max mem: 6508
Epoch: [8]  [103/104]  eta: 0:00:00  lr: 0.000010  loss: 53.0680 (55.5054)  loss_classifier: 0.7807 (1.3054)  loss_box_reg: 2.0435 (2.3042)  loss_objectness: 35.3156 (34.3672)  loss_rpn_box_reg: 15.0755 (17.5286)  time: 0.8918  data: 0.0191  max mem: 6508
Epoch: [8] Total time: 0:01:35 (0.9214 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:34  model_time: 0.4613 (0.4613)  evaluator_time: 0.0122 (0.0122)  time: 1.3162  data: 0.8309  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.3731 (0.3763)  evaluator_time: 0.0047 (0.0054)  time: 0.4072  data: 0.0207  max mem: 6508
Test: Total time: 0:00:11 (0.4443 s / it)
Averaged stats: model_time: 0.3731 (0.3763)  evaluator_time: 0.0047 (0.0054)
Accumulating evaluation results...
DONE (t=0.05s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
Epoch: [9]  [  0/104]  eta: 0:03:19  lr: 0.000001  loss: 55.8726 (55.8726)  loss_classifier: 0.9155 (0.9155)  loss_box_reg: 0.7995 (0.7995)  loss_objectness: 31.7795 (31.7795)  loss_rpn_box_reg: 22.3781 (22.3781)  time: 1.9175  data: 0.9353  max mem: 6508
Epoch: [9]  [ 10/104]  eta: 0:01:35  lr: 0.000001  loss: 36.0210 (36.4639)  loss_classifier: 0.7631 (0.9774)  loss_box_reg: 0.7818 (1.5536)  loss_objectness: 24.9991 (24.2409)  loss_rpn_box_reg: 5.7209 (9.6920)  time: 1.0133  data: 0.0998  max mem: 6508
Epoch: [9]  [ 20/104]  eta: 0:01:22  lr: 0.000001  loss: 36.0210 (46.6495)  loss_classifier: 0.5396 (0.9421)  loss_box_reg: 0.4097 (1.3542)  loss_objectness: 24.9991 (30.2645)  loss_rpn_box_reg: 5.7209 (14.0887)  time: 0.9318  data: 0.0175  max mem: 6508
Epoch: [9]  [ 30/104]  eta: 0:01:11  lr: 0.000001  loss: 47.9818 (50.8850)  loss_classifier: 0.5169 (1.0119)  loss_box_reg: 0.7757 (1.6858)  loss_objectness: 33.5256 (32.1530)  loss_rpn_box_reg: 10.3965 (16.0343)  time: 0.9348  data: 0.0207  max mem: 6508
Epoch: [9]  [ 40/104]  eta: 0:01:00  lr: 0.000001  loss: 56.1572 (58.4918)  loss_classifier: 0.6345 (1.1533)  loss_box_reg: 1.4540 (2.2561)  loss_objectness: 39.3350 (36.7060)  loss_rpn_box_reg: 10.8813 (18.3764)  time: 0.9227  data: 0.0226  max mem: 6508
Epoch: [9]  [ 50/104]  eta: 0:00:50  lr: 0.000001  loss: 40.9915 (58.5107)  loss_classifier: 1.1223 (1.3139)  loss_box_reg: 1.5430 (2.4593)  loss_objectness: 26.8529 (36.6794)  loss_rpn_box_reg: 7.0965 (18.0581)  time: 0.8986  data: 0.0208  max mem: 6508
Epoch: [9]  [ 60/104]  eta: 0:00:40  lr: 0.000001  loss: 40.7521 (59.0624)  loss_classifier: 0.7274 (1.2571)  loss_box_reg: 2.0573 (2.4919)  loss_objectness: 25.3522 (36.6843)  loss_rpn_box_reg: 6.5220 (18.6290)  time: 0.8901  data: 0.0210  max mem: 6508
Epoch: [9]  [ 70/104]  eta: 0:00:31  lr: 0.000001  loss: 41.9558 (57.6186)  loss_classifier: 0.5666 (1.2584)  loss_box_reg: 1.7006 (2.4362)  loss_objectness: 29.7488 (35.7220)  loss_rpn_box_reg: 6.8678 (18.2021)  time: 0.8941  data: 0.0224  max mem: 6508
Epoch: [9]  [ 80/104]  eta: 0:00:22  lr: 0.000001  loss: 38.1649 (56.1024)  loss_classifier: 0.9652 (1.2393)  loss_box_reg: 1.7006 (2.3581)  loss_objectness: 29.1210 (35.0864)  loss_rpn_box_reg: 7.8529 (17.4186)  time: 0.8848  data: 0.0208  max mem: 6508
Epoch: [9]  [ 90/104]  eta: 0:00:12  lr: 0.000001  loss: 39.4567 (55.4460)  loss_classifier: 0.6309 (1.1702)  loss_box_reg: 0.5337 (2.2205)  loss_objectness: 29.2802 (34.6866)  loss_rpn_box_reg: 9.7066 (17.3688)  time: 0.8894  data: 0.0209  max mem: 6508
Epoch: [9]  [100/104]  eta: 0:00:03  lr: 0.000001  loss: 46.5877 (55.7092)  loss_classifier: 0.5047 (1.2301)  loss_box_reg: 0.5314 (2.2278)  loss_objectness: 24.7870 (34.5281)  loss_rpn_box_reg: 12.9687 (17.7232)  time: 0.8960  data: 0.0205  max mem: 6508
Epoch: [9]  [103/104]  eta: 0:00:00  lr: 0.000001  loss: 45.1522 (54.8422)  loss_classifier: 0.5047 (1.2155)  loss_box_reg: 0.5314 (2.2548)  loss_objectness: 22.5050 (33.9452)  loss_rpn_box_reg: 10.2175 (17.4267)  time: 0.8943  data: 0.0194  max mem: 6508
Epoch: [9] Total time: 0:01:35 (0.9157 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:32  model_time: 0.5010 (0.5010)  evaluator_time: 0.0206 (0.0206)  time: 1.2450  data: 0.6964  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.3725 (0.3774)  evaluator_time: 0.0049 (0.0061)  time: 0.4074  data: 0.0213  max mem: 6508
Test: Total time: 0:00:11 (0.4435 s / it)
Averaged stats: model_time: 0.3725 (0.3774)  evaluator_time: 0.0049 (0.0061)
Accumulating evaluation results...
DONE (t=0.06s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
Epoch: [10]  [  0/104]  eta: 0:03:15  lr: 0.000001  loss: 136.4492 (136.4492)  loss_classifier: 0.7131 (0.7131)  loss_box_reg: 1.0747 (1.0747)  loss_objectness: 74.9324 (74.9324)  loss_rpn_box_reg: 59.7291 (59.7291)  time: 1.8833  data: 0.9262  max mem: 6508
Epoch: [10]  [ 10/104]  eta: 0:01:35  lr: 0.000001  loss: 33.3319 (60.0367)  loss_classifier: 1.4028 (1.6601)  loss_box_reg: 2.5008 (2.2897)  loss_objectness: 25.7583 (35.5108)  loss_rpn_box_reg: 6.7655 (20.5761)  time: 1.0110  data: 0.0995  max mem: 6508
Epoch: [10]  [ 20/104]  eta: 0:01:21  lr: 0.000001  loss: 33.3319 (52.6729)  loss_classifier: 0.7580 (1.3099)  loss_box_reg: 1.2173 (2.0461)  loss_objectness: 24.6153 (31.5986)  loss_rpn_box_reg: 7.6742 (17.7184)  time: 0.9304  data: 0.0183  max mem: 6508
Epoch: [10]  [ 30/104]  eta: 0:01:11  lr: 0.000001  loss: 37.1256 (56.9344)  loss_classifier: 0.6061 (1.1171)  loss_box_reg: 0.7630 (1.6936)  loss_objectness: 24.8843 (34.7807)  loss_rpn_box_reg: 11.1457 (19.3429)  time: 0.9365  data: 0.0201  max mem: 6508
Epoch: [10]  [ 40/104]  eta: 0:01:00  lr: 0.000001  loss: 30.8479 (54.6998)  loss_classifier: 0.5191 (1.1215)  loss_box_reg: 1.5854 (1.9849)  loss_objectness: 24.2222 (33.4807)  loss_rpn_box_reg: 8.9800 (18.1127)  time: 0.9287  data: 0.0233  max mem: 6508
Epoch: [10]  [ 50/104]  eta: 0:00:50  lr: 0.000001  loss: 50.0852 (59.1014)  loss_classifier: 0.4484 (0.9894)  loss_box_reg: 1.1914 (1.7593)  loss_objectness: 34.0017 (36.5236)  loss_rpn_box_reg: 9.6337 (19.8292)  time: 0.9103  data: 0.0237  max mem: 6508
Epoch: [10]  [ 60/104]  eta: 0:00:41  lr: 0.000001  loss: 40.0831 (56.1553)  loss_classifier: 0.4169 (1.0080)  loss_box_reg: 0.9465 (1.8049)  loss_objectness: 31.6625 (35.0517)  loss_rpn_box_reg: 11.9654 (18.2907)  time: 0.8914  data: 0.0217  max mem: 6508
Epoch: [10]  [ 70/104]  eta: 0:00:31  lr: 0.000001  loss: 40.0831 (57.7541)  loss_classifier: 0.6742 (1.1186)  loss_box_reg: 1.9267 (2.2065)  loss_objectness: 26.0857 (35.8218)  loss_rpn_box_reg: 11.1730 (18.6072)  time: 0.8808  data: 0.0212  max mem: 6508
Epoch: [10]  [ 80/104]  eta: 0:00:22  lr: 0.000001  loss: 43.0268 (56.7424)  loss_classifier: 0.7473 (1.0691)  loss_box_reg: 1.3503 (2.1036)  loss_objectness: 29.3088 (35.4298)  loss_rpn_box_reg: 10.3651 (18.1399)  time: 0.8871  data: 0.0218  max mem: 6508
Epoch: [10]  [ 90/104]  eta: 0:00:12  lr: 0.000001  loss: 37.6172 (55.5382)  loss_classifier: 0.6455 (1.0569)  loss_box_reg: 0.8860 (2.0527)  loss_objectness: 25.9196 (34.2333)  loss_rpn_box_reg: 8.6694 (18.1952)  time: 0.8944  data: 0.0222  max mem: 6508
Epoch: [10]  [100/104]  eta: 0:00:03  lr: 0.000001  loss: 34.2634 (55.1000)  loss_classifier: 0.7205 (1.1929)  loss_box_reg: 1.0591 (2.2203)  loss_objectness: 25.9196 (33.9396)  loss_rpn_box_reg: 8.3527 (17.7472)  time: 0.8922  data: 0.0199  max mem: 6508
Epoch: [10]  [103/104]  eta: 0:00:00  lr: 0.000001  loss: 38.6925 (54.6933)  loss_classifier: 1.0530 (1.1821)  loss_box_reg: 2.4034 (2.2524)  loss_objectness: 26.8310 (33.7915)  loss_rpn_box_reg: 8.3527 (17.4673)  time: 0.8956  data: 0.0196  max mem: 6508
Epoch: [10] Total time: 0:01:35 (0.9175 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:47  model_time: 0.6206 (0.6206)  evaluator_time: 0.0600 (0.0600)  time: 1.8325  data: 1.0885  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.3731 (0.3831)  evaluator_time: 0.0041 (0.0069)  time: 0.4015  data: 0.0178  max mem: 6508
Test: Total time: 0:00:12 (0.4636 s / it)
Averaged stats: model_time: 0.3731 (0.3831)  evaluator_time: 0.0041 (0.0069)
Accumulating evaluation results...
DONE (t=0.06s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
Epoch: [11]  [  0/104]  eta: 0:03:00  lr: 0.000001  loss: 52.8316 (52.8316)  loss_classifier: 0.1815 (0.1815)  loss_box_reg: 0.4922 (0.4922)  loss_objectness: 30.7382 (30.7382)  loss_rpn_box_reg: 21.4197 (21.4197)  time: 1.7318  data: 0.5542  max mem: 6508
Epoch: [11]  [ 10/104]  eta: 0:01:34  lr: 0.000001  loss: 52.8316 (57.8617)  loss_classifier: 0.5240 (0.6211)  loss_box_reg: 0.9764 (1.1655)  loss_objectness: 33.1904 (37.8400)  loss_rpn_box_reg: 15.1997 (18.2351)  time: 1.0077  data: 0.0694  max mem: 6508
Epoch: [11]  [ 20/104]  eta: 0:01:21  lr: 0.000001  loss: 43.8346 (53.9513)  loss_classifier: 0.5493 (1.0243)  loss_box_reg: 1.1809 (1.9616)  loss_objectness: 29.4245 (35.0679)  loss_rpn_box_reg: 9.4829 (15.8975)  time: 0.9359  data: 0.0206  max mem: 6508
Epoch: [11]  [ 30/104]  eta: 0:01:10  lr: 0.000001  loss: 37.5880 (53.6489)  loss_classifier: 0.9131 (1.1859)  loss_box_reg: 1.4377 (2.3371)  loss_objectness: 24.8980 (34.1686)  loss_rpn_box_reg: 8.3418 (15.9573)  time: 0.9246  data: 0.0194  max mem: 6508
Epoch: [11]  [ 40/104]  eta: 0:01:00  lr: 0.000001  loss: 37.4579 (51.5240)  loss_classifier: 1.0893 (1.1295)  loss_box_reg: 1.4377 (2.1583)  loss_objectness: 24.4681 (32.9490)  loss_rpn_box_reg: 8.4908 (15.2872)  time: 0.9086  data: 0.0196  max mem: 6508
Epoch: [11]  [ 50/104]  eta: 0:00:50  lr: 0.000001  loss: 40.1210 (55.8891)  loss_classifier: 0.6040 (1.0525)  loss_box_reg: 1.2593 (2.0721)  loss_objectness: 31.2911 (35.8685)  loss_rpn_box_reg: 8.4908 (16.8960)  time: 0.8972  data: 0.0209  max mem: 6508
Epoch: [11]  [ 60/104]  eta: 0:00:40  lr: 0.000001  loss: 42.8720 (54.6735)  loss_classifier: 0.5619 (0.9888)  loss_box_reg: 1.2123 (1.9293)  loss_objectness: 33.0219 (34.9984)  loss_rpn_box_reg: 13.5257 (16.7570)  time: 0.8877  data: 0.0203  max mem: 6508
Epoch: [11]  [ 70/104]  eta: 0:00:31  lr: 0.000001  loss: 47.5396 (54.0339)  loss_classifier: 0.5431 (0.9441)  loss_box_reg: 0.9506 (1.9152)  loss_objectness: 28.7603 (34.4323)  loss_rpn_box_reg: 14.2710 (16.7423)  time: 0.8907  data: 0.0220  max mem: 6508
Epoch: [11]  [ 80/104]  eta: 0:00:21  lr: 0.000001  loss: 36.6762 (55.7692)  loss_classifier: 0.7054 (1.0968)  loss_box_reg: 1.4827 (2.0605)  loss_objectness: 23.4399 (35.2508)  loss_rpn_box_reg: 6.7993 (17.3611)  time: 0.8920  data: 0.0228  max mem: 6508
Epoch: [11]  [ 90/104]  eta: 0:00:12  lr: 0.000001  loss: 36.6762 (54.3036)  loss_classifier: 0.7671 (1.0819)  loss_box_reg: 1.1549 (1.9902)  loss_objectness: 23.1004 (34.2691)  loss_rpn_box_reg: 6.8313 (16.9623)  time: 0.8888  data: 0.0204  max mem: 6508
Epoch: [11]  [100/104]  eta: 0:00:03  lr: 0.000001  loss: 38.1807 (54.7137)  loss_classifier: 0.7514 (1.1458)  loss_box_reg: 0.8284 (2.1847)  loss_objectness: 23.5308 (33.9078)  loss_rpn_box_reg: 13.1102 (17.4754)  time: 0.8988  data: 0.0205  max mem: 6508
Epoch: [11]  [103/104]  eta: 0:00:00  lr: 0.000001  loss: 41.9239 (54.5535)  loss_classifier: 0.7198 (1.1400)  loss_box_reg: 0.8284 (2.1824)  loss_objectness: 27.0699 (33.7869)  loss_rpn_box_reg: 13.1102 (17.4442)  time: 0.9017  data: 0.0202  max mem: 6508
Epoch: [11] Total time: 0:01:35 (0.9137 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:29  model_time: 0.4388 (0.4388)  evaluator_time: 0.0286 (0.0286)  time: 1.1380  data: 0.6646  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.3735 (0.3746)  evaluator_time: 0.0049 (0.0061)  time: 0.4048  data: 0.0209  max mem: 6508
Test: Total time: 0:00:11 (0.4384 s / it)
Averaged stats: model_time: 0.3735 (0.3746)  evaluator_time: 0.0049 (0.0061)
Accumulating evaluation results...
DONE (t=0.06s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
Epoch: [12]  [  0/104]  eta: 0:02:59  lr: 0.000000  loss: 22.1330 (22.1330)  loss_classifier: 0.2661 (0.2661)  loss_box_reg: 0.2869 (0.2869)  loss_objectness: 16.8623 (16.8623)  loss_rpn_box_reg: 4.7176 (4.7176)  time: 1.7247  data: 0.7689  max mem: 6508
Epoch: [12]  [ 10/104]  eta: 0:01:33  lr: 0.000000  loss: 41.8287 (56.1894)  loss_classifier: 0.4315 (1.0307)  loss_box_reg: 0.7968 (1.7654)  loss_objectness: 28.2230 (36.2898)  loss_rpn_box_reg: 7.5126 (17.1035)  time: 0.9989  data: 0.0867  max mem: 6508
Epoch: [12]  [ 20/104]  eta: 0:01:21  lr: 0.000000  loss: 32.3787 (46.6581)  loss_classifier: 0.4703 (0.8296)  loss_box_reg: 0.9784 (1.4492)  loss_objectness: 22.4661 (29.9172)  loss_rpn_box_reg: 7.2703 (14.4621)  time: 0.9285  data: 0.0205  max mem: 6508
Epoch: [12]  [ 30/104]  eta: 0:01:10  lr: 0.000000  loss: 30.8075 (51.9045)  loss_classifier: 0.6102 (0.9255)  loss_box_reg: 1.3047 (1.7918)  loss_objectness: 22.5891 (33.4390)  loss_rpn_box_reg: 7.9838 (15.7482)  time: 0.9170  data: 0.0207  max mem: 6508
Epoch: [12]  [ 40/104]  eta: 0:00:59  lr: 0.000000  loss: 36.1320 (51.0136)  loss_classifier: 0.5346 (0.9110)  loss_box_reg: 0.4441 (1.6896)  loss_objectness: 24.0349 (32.7051)  loss_rpn_box_reg: 8.2391 (15.7079)  time: 0.9038  data: 0.0200  max mem: 6508
Epoch: [12]  [ 50/104]  eta: 0:00:50  lr: 0.000000  loss: 38.1420 (50.6265)  loss_classifier: 0.3968 (0.8540)  loss_box_reg: 0.6667 (1.6038)  loss_objectness: 23.1181 (32.4891)  loss_rpn_box_reg: 9.0442 (15.6795)  time: 0.8978  data: 0.0205  max mem: 6508
Epoch: [12]  [ 60/104]  eta: 0:00:40  lr: 0.000000  loss: 47.5172 (51.1805)  loss_classifier: 0.4551 (0.8505)  loss_box_reg: 0.6767 (1.6152)  loss_objectness: 23.1181 (32.7655)  loss_rpn_box_reg: 16.0064 (15.9494)  time: 0.8884  data: 0.0198  max mem: 6508
Epoch: [12]  [ 70/104]  eta: 0:00:31  lr: 0.000000  loss: 39.2870 (50.6416)  loss_classifier: 0.6406 (0.8506)  loss_box_reg: 0.7314 (1.6614)  loss_objectness: 28.4020 (32.2610)  loss_rpn_box_reg: 9.2286 (15.8686)  time: 0.8869  data: 0.0220  max mem: 6508
Epoch: [12]  [ 80/104]  eta: 0:00:21  lr: 0.000000  loss: 39.6601 (51.5920)  loss_classifier: 0.7210 (0.9401)  loss_box_reg: 2.0792 (1.8347)  loss_objectness: 28.3379 (32.6720)  loss_rpn_box_reg: 9.5574 (16.1453)  time: 0.8899  data: 0.0229  max mem: 6508
Epoch: [12]  [ 90/104]  eta: 0:00:12  lr: 0.000000  loss: 44.6805 (53.8990)  loss_classifier: 0.7210 (1.0102)  loss_box_reg: 1.5198 (1.8423)  loss_objectness: 28.3379 (33.5128)  loss_rpn_box_reg: 10.2014 (17.5338)  time: 0.8901  data: 0.0208  max mem: 6508
Epoch: [12]  [100/104]  eta: 0:00:03  lr: 0.000000  loss: 42.8190 (53.3576)  loss_classifier: 0.4876 (1.0451)  loss_box_reg: 0.6095 (1.9236)  loss_objectness: 29.2609 (33.2374)  loss_rpn_box_reg: 9.7693 (17.1515)  time: 0.8921  data: 0.0200  max mem: 6508
Epoch: [12]  [103/104]  eta: 0:00:00  lr: 0.000000  loss: 42.8190 (53.7609)  loss_classifier: 0.7151 (1.0634)  loss_box_reg: 0.6377 (1.9176)  loss_objectness: 29.2609 (33.3749)  loss_rpn_box_reg: 9.6603 (17.4049)  time: 0.8897  data: 0.0189  max mem: 6508
Epoch: [12] Total time: 0:01:34 (0.9093 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:31  model_time: 0.4823 (0.4823)  evaluator_time: 0.0238 (0.0238)  time: 1.2010  data: 0.6737  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.3699 (0.3739)  evaluator_time: 0.0054 (0.0063)  time: 0.4055  data: 0.0223  max mem: 6508
Test: Total time: 0:00:11 (0.4382 s / it)
Averaged stats: model_time: 0.3699 (0.3739)  evaluator_time: 0.0054 (0.0063)
Accumulating evaluation results...
DONE (t=0.06s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
Epoch: [13]  [  0/104]  eta: 0:03:01  lr: 0.000000  loss: 44.5011 (44.5011)  loss_classifier: 2.7536 (2.7536)  loss_box_reg: 10.0774 (10.0774)  loss_objectness: 26.2172 (26.2172)  loss_rpn_box_reg: 5.4529 (5.4529)  time: 1.7486  data: 0.8313  max mem: 6508
Epoch: [13]  [ 10/104]  eta: 0:01:33  lr: 0.000000  loss: 44.2257 (53.2623)  loss_classifier: 0.5256 (1.0915)  loss_box_reg: 1.6258 (2.8355)  loss_objectness: 27.6522 (35.1720)  loss_rpn_box_reg: 8.7917 (14.1633)  time: 0.9956  data: 0.0903  max mem: 6508
Epoch: [13]  [ 20/104]  eta: 0:01:21  lr: 0.000000  loss: 44.2257 (56.4916)  loss_classifier: 0.7182 (1.4858)  loss_box_reg: 1.5482 (2.8480)  loss_objectness: 29.7883 (36.3452)  loss_rpn_box_reg: 12.7240 (15.8126)  time: 0.9343  data: 0.0204  max mem: 6508
Epoch: [13]  [ 30/104]  eta: 0:01:11  lr: 0.000000  loss: 39.6630 (55.2804)  loss_classifier: 0.7264 (1.1933)  loss_box_reg: 0.8544 (2.3042)  loss_objectness: 26.7828 (35.6702)  loss_rpn_box_reg: 14.2219 (16.1126)  time: 0.9396  data: 0.0246  max mem: 6508
Epoch: [13]  [ 40/104]  eta: 0:01:00  lr: 0.000000  loss: 39.6630 (52.4549)  loss_classifier: 0.5614 (1.1145)  loss_box_reg: 0.7318 (2.2849)  loss_objectness: 26.6561 (33.7647)  loss_rpn_box_reg: 10.6641 (15.2909)  time: 0.9201  data: 0.0227  max mem: 6508
Epoch: [13]  [ 50/104]  eta: 0:00:50  lr: 0.000000  loss: 37.3944 (51.5612)  loss_classifier: 0.7112 (1.1365)  loss_box_reg: 1.7637 (2.2915)  loss_objectness: 24.0970 (32.3410)  loss_rpn_box_reg: 9.6952 (15.7922)  time: 0.9014  data: 0.0206  max mem: 6508
Epoch: [13]  [ 60/104]  eta: 0:00:40  lr: 0.000000  loss: 37.3944 (54.3545)  loss_classifier: 0.8154 (1.0958)  loss_box_reg: 1.9316 (2.2330)  loss_objectness: 24.1391 (33.7685)  loss_rpn_box_reg: 9.6952 (17.2571)  time: 0.8955  data: 0.0210  max mem: 6508
Epoch: [13]  [ 70/104]  eta: 0:00:31  lr: 0.000000  loss: 48.4363 (55.8690)  loss_classifier: 0.8154 (1.3169)  loss_box_reg: 1.5628 (2.5893)  loss_objectness: 28.8844 (34.0468)  loss_rpn_box_reg: 10.0771 (17.9160)  time: 0.8865  data: 0.0207  max mem: 6508
Epoch: [13]  [ 80/104]  eta: 0:00:22  lr: 0.000000  loss: 42.7073 (56.7093)  loss_classifier: 0.7326 (1.2523)  loss_box_reg: 1.3432 (2.4407)  loss_objectness: 28.8844 (34.7017)  loss_rpn_box_reg: 11.1174 (18.3146)  time: 0.8817  data: 0.0203  max mem: 6508
Epoch: [13]  [ 90/104]  eta: 0:00:12  lr: 0.000000  loss: 35.5755 (55.1026)  loss_classifier: 0.7326 (1.2471)  loss_box_reg: 1.3305 (2.4795)  loss_objectness: 24.4649 (33.9262)  loss_rpn_box_reg: 6.0084 (17.4499)  time: 0.8874  data: 0.0212  max mem: 6508
Epoch: [13]  [100/104]  eta: 0:00:03  lr: 0.000000  loss: 38.7166 (54.9869)  loss_classifier: 0.6136 (1.1830)  loss_box_reg: 0.5408 (2.3141)  loss_objectness: 25.9364 (33.8751)  loss_rpn_box_reg: 6.3024 (17.6147)  time: 0.8934  data: 0.0216  max mem: 6508
Epoch: [13]  [103/104]  eta: 0:00:00  lr: 0.000000  loss: 38.7166 (54.6361)  loss_classifier: 0.5983 (1.1788)  loss_box_reg: 0.5408 (2.3405)  loss_objectness: 25.9364 (33.6802)  loss_rpn_box_reg: 8.0078 (17.4366)  time: 0.8947  data: 0.0209  max mem: 6508
Epoch: [13] Total time: 0:01:35 (0.9147 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:38  model_time: 0.5600 (0.5600)  evaluator_time: 0.0904 (0.0904)  time: 1.4916  data: 0.8302  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.3702 (0.3784)  evaluator_time: 0.0042 (0.0199)  time: 0.4151  data: 0.0197  max mem: 6508
Test: Total time: 0:00:12 (0.4646 s / it)
Averaged stats: model_time: 0.3702 (0.3784)  evaluator_time: 0.0042 (0.0199)
Accumulating evaluation results...
DONE (t=0.06s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
Epoch: [14]  [  0/104]  eta: 0:02:57  lr: 0.000000  loss: 7.2644 (7.2644)  loss_classifier: 0.4406 (0.4406)  loss_box_reg: 1.2378 (1.2378)  loss_objectness: 4.4630 (4.4630)  loss_rpn_box_reg: 1.1230 (1.1230)  time: 1.7115  data: 0.7214  max mem: 6508
Epoch: [14]  [ 10/104]  eta: 0:01:32  lr: 0.000000  loss: 40.6111 (41.7780)  loss_classifier: 0.4624 (1.8352)  loss_box_reg: 1.9378 (3.0898)  loss_objectness: 30.6523 (27.2508)  loss_rpn_box_reg: 9.4268 (9.6022)  time: 0.9852  data: 0.0847  max mem: 6508
Epoch: [14]  [ 20/104]  eta: 0:01:20  lr: 0.000000  loss: 41.9498 (46.2010)  loss_classifier: 0.5975 (1.4174)  loss_box_reg: 1.6083 (2.8503)  loss_objectness: 31.3423 (29.6717)  loss_rpn_box_reg: 9.5961 (12.2616)  time: 0.9239  data: 0.0220  max mem: 6508
Epoch: [14]  [ 30/104]  eta: 0:01:10  lr: 0.000000  loss: 44.8585 (49.7299)  loss_classifier: 0.9981 (1.4701)  loss_box_reg: 1.9661 (3.1523)  loss_objectness: 29.2475 (30.8669)  loss_rpn_box_reg: 12.0252 (14.2407)  time: 0.9251  data: 0.0211  max mem: 6508
Epoch: [14]  [ 40/104]  eta: 0:00:59  lr: 0.000000  loss: 54.1296 (55.8704)  loss_classifier: 0.9576 (1.2918)  loss_box_reg: 1.7528 (2.7358)  loss_objectness: 33.1939 (34.6591)  loss_rpn_box_reg: 14.3342 (17.1836)  time: 0.9105  data: 0.0196  max mem: 6508
Epoch: [14]  [ 50/104]  eta: 0:00:50  lr: 0.000000  loss: 45.0768 (54.3786)  loss_classifier: 0.4658 (1.1630)  loss_box_reg: 0.6312 (2.4449)  loss_objectness: 29.7384 (34.0258)  loss_rpn_box_reg: 12.8331 (16.7449)  time: 0.9031  data: 0.0222  max mem: 6508
Epoch: [14]  [ 60/104]  eta: 0:00:40  lr: 0.000000  loss: 36.1728 (52.7044)  loss_classifier: 0.5545 (1.1902)  loss_box_reg: 1.5292 (2.4030)  loss_objectness: 26.3899 (33.0513)  loss_rpn_box_reg: 8.3081 (16.0599)  time: 0.8876  data: 0.0215  max mem: 6508
Epoch: [14]  [ 70/104]  eta: 0:00:31  lr: 0.000000  loss: 41.8349 (55.4212)  loss_classifier: 0.5278 (1.1009)  loss_box_reg: 0.6700 (2.2442)  loss_objectness: 26.3899 (34.4498)  loss_rpn_box_reg: 8.2428 (17.6263)  time: 0.8831  data: 0.0212  max mem: 6508
Epoch: [14]  [ 80/104]  eta: 0:00:21  lr: 0.000000  loss: 35.9192 (53.3110)  loss_classifier: 0.5948 (1.1048)  loss_box_reg: 0.8445 (2.2483)  loss_objectness: 23.4596 (33.2773)  loss_rpn_box_reg: 8.9517 (16.6807)  time: 0.8901  data: 0.0232  max mem: 6508
Epoch: [14]  [ 90/104]  eta: 0:00:12  lr: 0.000000  loss: 33.5233 (53.9477)  loss_classifier: 0.6733 (1.0888)  loss_box_reg: 0.9395 (2.1349)  loss_objectness: 20.6789 (33.4692)  loss_rpn_box_reg: 9.4732 (17.2548)  time: 0.8903  data: 0.0207  max mem: 6508
Epoch: [14]  [100/104]  eta: 0:00:03  lr: 0.000000  loss: 41.9988 (54.3623)  loss_classifier: 0.7239 (1.1340)  loss_box_reg: 0.9588 (2.1179)  loss_objectness: 27.4681 (33.7159)  loss_rpn_box_reg: 9.7129 (17.3945)  time: 0.9045  data: 0.0208  max mem: 6508
Epoch: [14]  [103/104]  eta: 0:00:00  lr: 0.000000  loss: 41.9988 (53.9387)  loss_classifier: 0.6733 (1.1094)  loss_box_reg: 0.8385 (2.0599)  loss_objectness: 27.8470 (33.3843)  loss_rpn_box_reg: 11.5158 (17.3851)  time: 0.9092  data: 0.0206  max mem: 6508
Epoch: [14] Total time: 0:01:34 (0.9127 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:32  model_time: 0.4477 (0.4477)  evaluator_time: 0.0099 (0.0099)  time: 1.2394  data: 0.7516  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.3715 (0.3737)  evaluator_time: 0.0051 (0.0055)  time: 0.4060  data: 0.0216  max mem: 6508
Test: Total time: 0:00:11 (0.4405 s / it)
Averaged stats: model_time: 0.3715 (0.3737)  evaluator_time: 0.0051 (0.0055)
Accumulating evaluation results...
DONE (t=0.06s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000

#save rmsprob
import pickle
Filename = "FRCNN3rmsprob.pkl"
# Define the file path where you want to save the model
filename = "/content/drive/MyDrive/dataset1/FRCNN3rmsprob.pkl"

# Save the model to the specified file path
torch.save(model.state_dict(), filename)
# Save the Modle to file in the current working directory
with open(Filename, 'wb') as file:
    pickle.dump(model, file)
# Load the Model back from file
with open(Filename, 'rb') as file:
    model = pickle.load(file)
model

FasterRCNN(
  (transform): GeneralizedRCNNTransform(
      Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
      Resize(min_size=(800,), max_size=1333, mode='bilinear')
  )
  (backbone): BackboneWithFPN(
    (body): IntermediateLayerGetter(
      (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)
      (bn1): FrozenBatchNorm2d(64, eps=0.0)
      (relu): ReLU(inplace=True)
      (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)
      (layer1): Sequential(
        (0): Bottleneck(
          (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(64, eps=0.0)
          (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(64, eps=0.0)
          (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(256, eps=0.0)
          (relu): ReLU(inplace=True)
          (downsample): Sequential(
            (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
            (1): FrozenBatchNorm2d(256, eps=0.0)
          )
        )
        (1): Bottleneck(
          (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(64, eps=0.0)
          (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(64, eps=0.0)
          (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(256, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (2): Bottleneck(
          (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(64, eps=0.0)
          (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(64, eps=0.0)
          (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(256, eps=0.0)
          (relu): ReLU(inplace=True)
        )
      )
      (layer2): Sequential(
        (0): Bottleneck(
          (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(128, eps=0.0)
          (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(128, eps=0.0)
          (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(512, eps=0.0)
          (relu): ReLU(inplace=True)
          (downsample): Sequential(
            (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)
            (1): FrozenBatchNorm2d(512, eps=0.0)
          )
        )
        (1): Bottleneck(
          (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(128, eps=0.0)
          (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(128, eps=0.0)
          (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(512, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (2): Bottleneck(
          (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(128, eps=0.0)
          (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(128, eps=0.0)
          (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(512, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (3): Bottleneck(
          (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(128, eps=0.0)
          (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(128, eps=0.0)
          (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(512, eps=0.0)
          (relu): ReLU(inplace=True)
        )
      )
      (layer3): Sequential(
        (0): Bottleneck(
          (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
          (downsample): Sequential(
            (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False)
            (1): FrozenBatchNorm2d(1024, eps=0.0)
          )
        )
        (1): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (2): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (3): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (4): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (5): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
        )
      )
      (layer4): Sequential(
        (0): Bottleneck(
          (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(512, eps=0.0)
          (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(512, eps=0.0)
          (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(2048, eps=0.0)
          (relu): ReLU(inplace=True)
          (downsample): Sequential(
            (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False)
            (1): FrozenBatchNorm2d(2048, eps=0.0)
          )
        )
        (1): Bottleneck(
          (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(512, eps=0.0)
          (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(512, eps=0.0)
          (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(2048, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (2): Bottleneck(
          (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(512, eps=0.0)
          (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(512, eps=0.0)
          (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(2048, eps=0.0)
          (relu): ReLU(inplace=True)
        )
      )
    )
    (fpn): FeaturePyramidNetwork(
      (inner_blocks): ModuleList(
        (0): Conv2dNormActivation(
          (0): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1))
        )
        (1): Conv2dNormActivation(
          (0): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1))
        )
        (2): Conv2dNormActivation(
          (0): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1))
        )
        (3): Conv2dNormActivation(
          (0): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1))
        )
      )
      (layer_blocks): ModuleList(
        (0-3): 4 x Conv2dNormActivation(
          (0): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
        )
      )
      (extra_blocks): LastLevelMaxPool()
    )
  )
  (rpn): RegionProposalNetwork(
    (anchor_generator): AnchorGenerator()
    (head): RPNHead(
      (conv): Sequential(
        (0): Conv2dNormActivation(
          (0): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
          (1): ReLU(inplace=True)
        )
      )
      (cls_logits): Conv2d(256, 3, kernel_size=(1, 1), stride=(1, 1))
      (bbox_pred): Conv2d(256, 12, kernel_size=(1, 1), stride=(1, 1))
    )
  )
  (roi_heads): RoIHeads(
    (box_roi_pool): MultiScaleRoIAlign(featmap_names=['0', '1', '2', '3'], output_size=(7, 7), sampling_ratio=2)
    (box_head): TwoMLPHead(
      (fc6): Linear(in_features=12544, out_features=1024, bias=True)
      (fc7): Linear(in_features=1024, out_features=1024, bias=True)
    )
    (box_predictor): FastRCNNPredictor(
      (cls_score): Linear(in_features=1024, out_features=11, bias=True)
      (bbox_pred): Linear(in_features=1024, out_features=44, bias=True)
    )
  )
)

#adelta
# to train on GPU if selected
device = torch.device('cuda') if torch.cuda.is_available() else torch.device('cpu')

# number of classes
num_classes = 11

# get the model using our helper function
model = get_object_detection_model(num_classes)

# move model to the right device
model.to(device)

# construct an optimizer
params = [p for p in model.parameters() if p.requires_grad]
optimizer = torch.optim.Adadelta(params, lr=0.001, rho=0.9, eps=1e-06, weight_decay=0.0005)

# and a learning rate scheduler which decreases the learning rate by
# 10x every 3 epochs
lr_scheduler = torch.optim.lr_scheduler.StepLR(optimizer,
                                               step_size=3,
                                               gamma=0.1)

# training for 8 epochs # adekta
num_epochs = 15

for epoch in range(num_epochs):
    # training for one epoch
    train_one_epoch(model, optimizer, data_loader, device, epoch, print_freq=10)
    # update the learning rate
    lr_scheduler.step()
    # evaluate on the test dataset
    evaluate(model, data_loader_test, device=device)

Epoch: [0]  [  0/104]  eta: 0:03:49  lr: 0.000011  loss: 4.0759 (4.0759)  loss_classifier: 2.5687 (2.5687)  loss_box_reg: 0.2849 (0.2849)  loss_objectness: 1.1855 (1.1855)  loss_rpn_box_reg: 0.0368 (0.0368)  time: 2.2106  data: 0.9448  max mem: 6508
Epoch: [0]  [ 10/104]  eta: 0:02:00  lr: 0.000108  loss: 3.3600 (3.3749)  loss_classifier: 2.6055 (2.6007)  loss_box_reg: 0.2849 (0.2693)  loss_objectness: 0.3081 (0.4810)  loss_rpn_box_reg: 0.0216 (0.0239)  time: 1.2857  data: 0.1197  max mem: 6508
Epoch: [0]  [ 20/104]  eta: 0:01:43  lr: 0.000205  loss: 3.1378 (3.2732)  loss_classifier: 2.5849 (2.5738)  loss_box_reg: 0.1922 (0.2402)  loss_objectness: 0.2341 (0.4338)  loss_rpn_box_reg: 0.0178 (0.0253)  time: 1.1830  data: 0.0391  max mem: 6508
Epoch: [0]  [ 30/104]  eta: 0:01:28  lr: 0.000302  loss: 2.9602 (3.1843)  loss_classifier: 2.5095 (2.5413)  loss_box_reg: 0.1580 (0.2168)  loss_objectness: 0.2463 (0.4034)  loss_rpn_box_reg: 0.0178 (0.0228)  time: 1.1476  data: 0.0378  max mem: 6508
Epoch: [0]  [ 40/104]  eta: 0:01:14  lr: 0.000399  loss: 2.9602 (3.1469)  loss_classifier: 2.4297 (2.4981)  loss_box_reg: 0.1765 (0.2324)  loss_objectness: 0.3229 (0.3927)  loss_rpn_box_reg: 0.0201 (0.0237)  time: 1.0834  data: 0.0269  max mem: 6508
Epoch: [0]  [ 50/104]  eta: 0:01:01  lr: 0.000496  loss: 2.8197 (3.0742)  loss_classifier: 2.2737 (2.4437)  loss_box_reg: 0.2803 (0.2405)  loss_objectness: 0.3229 (0.3664)  loss_rpn_box_reg: 0.0216 (0.0236)  time: 1.0403  data: 0.0201  max mem: 6508
Epoch: [0]  [ 60/104]  eta: 0:00:49  lr: 0.000593  loss: 2.6611 (2.9950)  loss_classifier: 2.1325 (2.3786)  loss_box_reg: 0.2608 (0.2478)  loss_objectness: 0.2339 (0.3454)  loss_rpn_box_reg: 0.0205 (0.0232)  time: 1.0377  data: 0.0210  max mem: 6508
Epoch: [0]  [ 70/104]  eta: 0:00:37  lr: 0.000690  loss: 2.4501 (2.8952)  loss_classifier: 1.9114 (2.3037)  loss_box_reg: 0.2562 (0.2480)  loss_objectness: 0.1492 (0.3212)  loss_rpn_box_reg: 0.0181 (0.0222)  time: 1.0399  data: 0.0207  max mem: 6508
Epoch: [0]  [ 80/104]  eta: 0:00:26  lr: 0.000787  loss: 2.0663 (2.7903)  loss_classifier: 1.7363 (2.2144)  loss_box_reg: 0.2166 (0.2499)  loss_objectness: 0.1413 (0.3041)  loss_rpn_box_reg: 0.0146 (0.0218)  time: 1.0462  data: 0.0195  max mem: 6508
Epoch: [0]  [ 90/104]  eta: 0:00:15  lr: 0.000884  loss: 1.8681 (2.6737)  loss_classifier: 1.3998 (2.1079)  loss_box_reg: 0.2123 (0.2480)  loss_objectness: 0.1271 (0.2961)  loss_rpn_box_reg: 0.0156 (0.0217)  time: 1.0609  data: 0.0198  max mem: 6508
Epoch: [0]  [100/104]  eta: 0:00:04  lr: 0.000981  loss: 1.6345 (2.5573)  loss_classifier: 1.1006 (1.9976)  loss_box_reg: 0.2507 (0.2551)  loss_objectness: 0.1404 (0.2830)  loss_rpn_box_reg: 0.0193 (0.0217)  time: 1.0691  data: 0.0198  max mem: 6508
Epoch: [0]  [103/104]  eta: 0:00:01  lr: 0.001000  loss: 1.6345 (2.5272)  loss_classifier: 1.0552 (1.9649)  loss_box_reg: 0.2609 (0.2581)  loss_objectness: 0.1482 (0.2819)  loss_rpn_box_reg: 0.0211 (0.0224)  time: 1.0684  data: 0.0193  max mem: 6508
Epoch: [0] Total time: 0:01:53 (1.0955 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:32  model_time: 0.5306 (0.5306)  evaluator_time: 0.0470 (0.0470)  time: 1.2572  data: 0.6638  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4439 (0.4468)  evaluator_time: 0.0204 (0.0226)  time: 0.4938  data: 0.0198  max mem: 6508
Test: Total time: 0:00:13 (0.5274 s / it)
Averaged stats: model_time: 0.4439 (0.4468)  evaluator_time: 0.0204 (0.0226)
Accumulating evaluation results...
DONE (t=0.34s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.001
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.002
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.002
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.001
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.011
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.016
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.014
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.006
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.008
Epoch: [1]  [  0/104]  eta: 0:06:03  lr: 0.001000  loss: 1.1351 (1.1351)  loss_classifier: 0.7006 (0.7006)  loss_box_reg: 0.2811 (0.2811)  loss_objectness: 0.1311 (0.1311)  loss_rpn_box_reg: 0.0222 (0.0222)  time: 3.4963  data: 1.7947  max mem: 6508
Epoch: [1]  [ 10/104]  eta: 0:02:07  lr: 0.001000  loss: 1.1702 (1.2361)  loss_classifier: 0.7006 (0.7172)  loss_box_reg: 0.3579 (0.3686)  loss_objectness: 0.0991 (0.1220)  loss_rpn_box_reg: 0.0222 (0.0283)  time: 1.3578  data: 0.1842  max mem: 6508
Epoch: [1]  [ 20/104]  eta: 0:01:44  lr: 0.001000  loss: 1.0701 (1.1490)  loss_classifier: 0.5925 (0.6332)  loss_box_reg: 0.3021 (0.3316)  loss_objectness: 0.0991 (0.1561)  loss_rpn_box_reg: 0.0203 (0.0282)  time: 1.1287  data: 0.0231  max mem: 6508
Epoch: [1]  [ 30/104]  eta: 0:01:28  lr: 0.001000  loss: 0.9767 (1.1083)  loss_classifier: 0.5344 (0.6057)  loss_box_reg: 0.3021 (0.3298)  loss_objectness: 0.1022 (0.1467)  loss_rpn_box_reg: 0.0179 (0.0262)  time: 1.1007  data: 0.0215  max mem: 6508
Epoch: [1]  [ 40/104]  eta: 0:01:14  lr: 0.001000  loss: 0.9272 (1.0977)  loss_classifier: 0.5014 (0.5869)  loss_box_reg: 0.2982 (0.3314)  loss_objectness: 0.1022 (0.1527)  loss_rpn_box_reg: 0.0179 (0.0267)  time: 1.0782  data: 0.0200  max mem: 6508
Epoch: [1]  [ 50/104]  eta: 0:01:01  lr: 0.001000  loss: 0.8314 (1.0514)  loss_classifier: 0.4376 (0.5612)  loss_box_reg: 0.2701 (0.3235)  loss_objectness: 0.1048 (0.1416)  loss_rpn_box_reg: 0.0159 (0.0251)  time: 1.0561  data: 0.0195  max mem: 6508
Epoch: [1]  [ 60/104]  eta: 0:00:49  lr: 0.001000  loss: 0.8384 (1.0255)  loss_classifier: 0.4454 (0.5453)  loss_box_reg: 0.2669 (0.3180)  loss_objectness: 0.1022 (0.1381)  loss_rpn_box_reg: 0.0159 (0.0241)  time: 1.0470  data: 0.0209  max mem: 6508
Epoch: [1]  [ 70/104]  eta: 0:00:37  lr: 0.001000  loss: 0.9086 (1.0151)  loss_classifier: 0.4537 (0.5380)  loss_box_reg: 0.2953 (0.3217)  loss_objectness: 0.0950 (0.1325)  loss_rpn_box_reg: 0.0155 (0.0230)  time: 1.0571  data: 0.0259  max mem: 6508
Epoch: [1]  [ 80/104]  eta: 0:00:26  lr: 0.001000  loss: 0.8851 (0.9949)  loss_classifier: 0.4499 (0.5273)  loss_box_reg: 0.2936 (0.3193)  loss_objectness: 0.0816 (0.1264)  loss_rpn_box_reg: 0.0130 (0.0219)  time: 1.0643  data: 0.0281  max mem: 6508
Epoch: [1]  [ 90/104]  eta: 0:00:15  lr: 0.001000  loss: 0.8026 (0.9790)  loss_classifier: 0.4275 (0.5167)  loss_box_reg: 0.2936 (0.3164)  loss_objectness: 0.0677 (0.1242)  loss_rpn_box_reg: 0.0123 (0.0218)  time: 1.0691  data: 0.0244  max mem: 6508
Epoch: [1]  [100/104]  eta: 0:00:04  lr: 0.001000  loss: 0.9113 (0.9848)  loss_classifier: 0.4669 (0.5179)  loss_box_reg: 0.3101 (0.3238)  loss_objectness: 0.0627 (0.1213)  loss_rpn_box_reg: 0.0149 (0.0218)  time: 1.0735  data: 0.0212  max mem: 6508
Epoch: [1]  [103/104]  eta: 0:00:01  lr: 0.001000  loss: 0.9113 (0.9854)  loss_classifier: 0.4463 (0.5175)  loss_box_reg: 0.3214 (0.3253)  loss_objectness: 0.0627 (0.1205)  loss_rpn_box_reg: 0.0173 (0.0220)  time: 1.0693  data: 0.0209  max mem: 6508
Epoch: [1] Total time: 0:01:54 (1.1039 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:46  model_time: 0.6684 (0.6684)  evaluator_time: 0.0484 (0.0484)  time: 1.8000  data: 1.0774  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4410 (0.4497)  evaluator_time: 0.0165 (0.0178)  time: 0.4830  data: 0.0187  max mem: 6508
Test: Total time: 0:00:14 (0.5398 s / it)
Averaged stats: model_time: 0.4410 (0.4497)  evaluator_time: 0.0165 (0.0178)
Accumulating evaluation results...
DONE (t=0.26s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.011
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.030
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.005
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.018
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.006
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.006
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.041
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.051
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.107
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.038
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
Epoch: [2]  [  0/104]  eta: 0:04:23  lr: 0.001000  loss: 0.9163 (0.9163)  loss_classifier: 0.4803 (0.4803)  loss_box_reg: 0.3495 (0.3495)  loss_objectness: 0.0612 (0.0612)  loss_rpn_box_reg: 0.0254 (0.0254)  time: 2.5327  data: 1.2497  max mem: 6508
Epoch: [2]  [ 10/104]  eta: 0:01:54  lr: 0.001000  loss: 0.9163 (0.9454)  loss_classifier: 0.4786 (0.4850)  loss_box_reg: 0.3385 (0.3521)  loss_objectness: 0.0614 (0.0885)  loss_rpn_box_reg: 0.0140 (0.0198)  time: 1.2232  data: 0.1286  max mem: 6508
Epoch: [2]  [ 20/104]  eta: 0:01:38  lr: 0.001000  loss: 0.9086 (0.9703)  loss_classifier: 0.4786 (0.4987)  loss_box_reg: 0.3239 (0.3639)  loss_objectness: 0.0696 (0.0875)  loss_rpn_box_reg: 0.0162 (0.0201)  time: 1.1009  data: 0.0191  max mem: 6508
Epoch: [2]  [ 30/104]  eta: 0:01:24  lr: 0.001000  loss: 0.9709 (0.9801)  loss_classifier: 0.5130 (0.5027)  loss_box_reg: 0.3860 (0.3734)  loss_objectness: 0.0691 (0.0841)  loss_rpn_box_reg: 0.0168 (0.0199)  time: 1.1000  data: 0.0229  max mem: 6508
Epoch: [2]  [ 40/104]  eta: 0:01:12  lr: 0.001000  loss: 0.9518 (0.9552)  loss_classifier: 0.5024 (0.4898)  loss_box_reg: 0.3826 (0.3658)  loss_objectness: 0.0632 (0.0802)  loss_rpn_box_reg: 0.0146 (0.0194)  time: 1.0862  data: 0.0257  max mem: 6508
Epoch: [2]  [ 50/104]  eta: 0:01:00  lr: 0.001000  loss: 0.8419 (0.9578)  loss_classifier: 0.4381 (0.4930)  loss_box_reg: 0.3406 (0.3668)  loss_objectness: 0.0576 (0.0781)  loss_rpn_box_reg: 0.0143 (0.0199)  time: 1.0854  data: 0.0258  max mem: 6508
Epoch: [2]  [ 60/104]  eta: 0:00:48  lr: 0.001000  loss: 0.8895 (0.9859)  loss_classifier: 0.4577 (0.5066)  loss_box_reg: 0.3665 (0.3822)  loss_objectness: 0.0575 (0.0766)  loss_rpn_box_reg: 0.0192 (0.0205)  time: 1.0645  data: 0.0218  max mem: 6508
Epoch: [2]  [ 70/104]  eta: 0:00:37  lr: 0.001000  loss: 0.9628 (0.9828)  loss_classifier: 0.5025 (0.5034)  loss_box_reg: 0.3798 (0.3791)  loss_objectness: 0.0557 (0.0795)  loss_rpn_box_reg: 0.0189 (0.0208)  time: 1.0458  data: 0.0200  max mem: 6508
Epoch: [2]  [ 80/104]  eta: 0:00:26  lr: 0.001000  loss: 0.9628 (0.9918)  loss_classifier: 0.5025 (0.5083)  loss_box_reg: 0.3798 (0.3861)  loss_objectness: 0.0529 (0.0767)  loss_rpn_box_reg: 0.0128 (0.0207)  time: 1.0572  data: 0.0214  max mem: 6508
Epoch: [2]  [ 90/104]  eta: 0:00:15  lr: 0.001000  loss: 0.8832 (0.9863)  loss_classifier: 0.4909 (0.5062)  loss_box_reg: 0.3771 (0.3855)  loss_objectness: 0.0475 (0.0745)  loss_rpn_box_reg: 0.0130 (0.0202)  time: 1.0670  data: 0.0217  max mem: 6508
Epoch: [2]  [100/104]  eta: 0:00:04  lr: 0.001000  loss: 0.8006 (0.9714)  loss_classifier: 0.4076 (0.4977)  loss_box_reg: 0.3359 (0.3820)  loss_objectness: 0.0467 (0.0722)  loss_rpn_box_reg: 0.0130 (0.0196)  time: 1.0693  data: 0.0195  max mem: 6508
Epoch: [2]  [103/104]  eta: 0:00:01  lr: 0.001000  loss: 0.8006 (0.9672)  loss_classifier: 0.4076 (0.4962)  loss_box_reg: 0.3424 (0.3807)  loss_objectness: 0.0454 (0.0710)  loss_rpn_box_reg: 0.0102 (0.0193)  time: 1.0710  data: 0.0193  max mem: 6508
Epoch: [2] Total time: 0:01:53 (1.0908 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:33  model_time: 0.5147 (0.5147)  evaluator_time: 0.0170 (0.0170)  time: 1.2991  data: 0.7548  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4410 (0.4421)  evaluator_time: 0.0223 (0.0248)  time: 0.4912  data: 0.0194  max mem: 6508
Test: Total time: 0:00:13 (0.5259 s / it)
Averaged stats: model_time: 0.4410 (0.4421)  evaluator_time: 0.0223 (0.0248)
Accumulating evaluation results...
DONE (t=0.18s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.025
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.065
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.012
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.031
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.023
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.004
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.012
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.077
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.104
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.135
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.139
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.025
Epoch: [3]  [  0/104]  eta: 0:03:15  lr: 0.000100  loss: 1.0471 (1.0471)  loss_classifier: 0.5305 (0.5305)  loss_box_reg: 0.4053 (0.4053)  loss_objectness: 0.0849 (0.0849)  loss_rpn_box_reg: 0.0264 (0.0264)  time: 1.8781  data: 0.7191  max mem: 6508
Epoch: [3]  [ 10/104]  eta: 0:01:50  lr: 0.000100  loss: 1.0512 (1.1071)  loss_classifier: 0.5372 (0.5524)  loss_box_reg: 0.4420 (0.4525)  loss_objectness: 0.0744 (0.0766)  loss_rpn_box_reg: 0.0263 (0.0257)  time: 1.1776  data: 0.0861  max mem: 6508
Epoch: [3]  [ 20/104]  eta: 0:01:36  lr: 0.000100  loss: 1.0315 (1.0544)  loss_classifier: 0.5223 (0.5318)  loss_box_reg: 0.4315 (0.4377)  loss_objectness: 0.0517 (0.0633)  loss_rpn_box_reg: 0.0218 (0.0215)  time: 1.1065  data: 0.0210  max mem: 6508
Epoch: [3]  [ 30/104]  eta: 0:01:23  lr: 0.000100  loss: 0.9557 (1.0177)  loss_classifier: 0.4734 (0.5138)  loss_box_reg: 0.3947 (0.4212)  loss_objectness: 0.0451 (0.0626)  loss_rpn_box_reg: 0.0136 (0.0201)  time: 1.0928  data: 0.0195  max mem: 6508
Epoch: [3]  [ 40/104]  eta: 0:01:11  lr: 0.000100  loss: 0.9115 (0.9881)  loss_classifier: 0.4660 (0.5027)  loss_box_reg: 0.3792 (0.4083)  loss_objectness: 0.0462 (0.0583)  loss_rpn_box_reg: 0.0135 (0.0188)  time: 1.0741  data: 0.0208  max mem: 6508
Epoch: [3]  [ 50/104]  eta: 0:00:59  lr: 0.000100  loss: 0.9662 (0.9973)  loss_classifier: 0.5089 (0.5078)  loss_box_reg: 0.4089 (0.4124)  loss_objectness: 0.0467 (0.0587)  loss_rpn_box_reg: 0.0165 (0.0184)  time: 1.0588  data: 0.0212  max mem: 6508
Epoch: [3]  [ 60/104]  eta: 0:00:47  lr: 0.000100  loss: 0.9121 (0.9782)  loss_classifier: 0.4642 (0.4988)  loss_box_reg: 0.3875 (0.4065)  loss_objectness: 0.0473 (0.0553)  loss_rpn_box_reg: 0.0128 (0.0175)  time: 1.0466  data: 0.0205  max mem: 6508
Epoch: [3]  [ 70/104]  eta: 0:00:36  lr: 0.000100  loss: 0.8367 (0.9782)  loss_classifier: 0.4401 (0.4975)  loss_box_reg: 0.3690 (0.4061)  loss_objectness: 0.0472 (0.0564)  loss_rpn_box_reg: 0.0111 (0.0182)  time: 1.0455  data: 0.0202  max mem: 6508
Epoch: [3]  [ 80/104]  eta: 0:00:25  lr: 0.000100  loss: 1.0318 (0.9807)  loss_classifier: 0.5163 (0.4972)  loss_box_reg: 0.3980 (0.4057)  loss_objectness: 0.0539 (0.0591)  loss_rpn_box_reg: 0.0163 (0.0187)  time: 1.0587  data: 0.0215  max mem: 6508
Epoch: [3]  [ 90/104]  eta: 0:00:15  lr: 0.000100  loss: 0.9981 (0.9717)  loss_classifier: 0.4935 (0.4919)  loss_box_reg: 0.3643 (0.4011)  loss_objectness: 0.0574 (0.0605)  loss_rpn_box_reg: 0.0160 (0.0181)  time: 1.0741  data: 0.0221  max mem: 6508
Epoch: [3]  [100/104]  eta: 0:00:04  lr: 0.000100  loss: 0.9981 (0.9730)  loss_classifier: 0.5089 (0.4929)  loss_box_reg: 0.4189 (0.4013)  loss_objectness: 0.0501 (0.0604)  loss_rpn_box_reg: 0.0158 (0.0183)  time: 1.0747  data: 0.0209  max mem: 6508
Epoch: [3]  [103/104]  eta: 0:00:01  lr: 0.000100  loss: 1.0118 (0.9736)  loss_classifier: 0.5108 (0.4932)  loss_box_reg: 0.4203 (0.4021)  loss_objectness: 0.0478 (0.0599)  loss_rpn_box_reg: 0.0167 (0.0183)  time: 1.0769  data: 0.0216  max mem: 6508
Epoch: [3] Total time: 0:01:52 (1.0812 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:31  model_time: 0.5568 (0.5568)  evaluator_time: 0.0257 (0.0257)  time: 1.2150  data: 0.6123  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4423 (0.4446)  evaluator_time: 0.0232 (0.0265)  time: 0.4953  data: 0.0208  max mem: 6508
Test: Total time: 0:00:13 (0.5257 s / it)
Averaged stats: model_time: 0.4423 (0.4446)  evaluator_time: 0.0232 (0.0265)
Accumulating evaluation results...
DONE (t=0.18s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.026
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.069
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.012
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.032
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.025
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.004
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.013
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.082
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.114
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.152
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.147
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.025
Epoch: [4]  [  0/104]  eta: 0:03:14  lr: 0.000100  loss: 1.0600 (1.0600)  loss_classifier: 0.5243 (0.5243)  loss_box_reg: 0.4688 (0.4688)  loss_objectness: 0.0506 (0.0506)  loss_rpn_box_reg: 0.0161 (0.0161)  time: 1.8730  data: 0.6827  max mem: 6508
Epoch: [4]  [ 10/104]  eta: 0:01:49  lr: 0.000100  loss: 0.9999 (1.0436)  loss_classifier: 0.5243 (0.5329)  loss_box_reg: 0.4057 (0.4451)  loss_objectness: 0.0481 (0.0462)  loss_rpn_box_reg: 0.0161 (0.0193)  time: 1.1693  data: 0.0823  max mem: 6508
Epoch: [4]  [ 20/104]  eta: 0:01:35  lr: 0.000100  loss: 0.9513 (1.0134)  loss_classifier: 0.4843 (0.5122)  loss_box_reg: 0.3950 (0.4227)  loss_objectness: 0.0481 (0.0597)  loss_rpn_box_reg: 0.0160 (0.0187)  time: 1.1053  data: 0.0205  max mem: 6508
Epoch: [4]  [ 30/104]  eta: 0:01:23  lr: 0.000100  loss: 0.9513 (0.9962)  loss_classifier: 0.4843 (0.5028)  loss_box_reg: 0.3740 (0.4104)  loss_objectness: 0.0565 (0.0631)  loss_rpn_box_reg: 0.0195 (0.0198)  time: 1.1017  data: 0.0205  max mem: 6508
Epoch: [4]  [ 40/104]  eta: 0:01:11  lr: 0.000100  loss: 0.9775 (0.9923)  loss_classifier: 0.4883 (0.5035)  loss_box_reg: 0.4030 (0.4087)  loss_objectness: 0.0565 (0.0606)  loss_rpn_box_reg: 0.0212 (0.0195)  time: 1.0805  data: 0.0224  max mem: 6508
Epoch: [4]  [ 50/104]  eta: 0:00:59  lr: 0.000100  loss: 0.8397 (0.9709)  loss_classifier: 0.4366 (0.4946)  loss_box_reg: 0.3437 (0.4013)  loss_objectness: 0.0421 (0.0570)  loss_rpn_box_reg: 0.0129 (0.0181)  time: 1.0649  data: 0.0245  max mem: 6508
Epoch: [4]  [ 60/104]  eta: 0:00:48  lr: 0.000100  loss: 0.8397 (0.9653)  loss_classifier: 0.4268 (0.4909)  loss_box_reg: 0.3625 (0.4005)  loss_objectness: 0.0399 (0.0568)  loss_rpn_box_reg: 0.0093 (0.0171)  time: 1.0545  data: 0.0240  max mem: 6508
Epoch: [4]  [ 70/104]  eta: 0:00:36  lr: 0.000100  loss: 0.8745 (0.9758)  loss_classifier: 0.4384 (0.4945)  loss_box_reg: 0.3625 (0.4033)  loss_objectness: 0.0562 (0.0594)  loss_rpn_box_reg: 0.0174 (0.0186)  time: 1.0504  data: 0.0209  max mem: 6508
Epoch: [4]  [ 80/104]  eta: 0:00:26  lr: 0.000100  loss: 0.8969 (0.9721)  loss_classifier: 0.4830 (0.4912)  loss_box_reg: 0.3493 (0.4027)  loss_objectness: 0.0634 (0.0593)  loss_rpn_box_reg: 0.0193 (0.0190)  time: 1.0562  data: 0.0203  max mem: 6508
Epoch: [4]  [ 90/104]  eta: 0:00:15  lr: 0.000100  loss: 0.8969 (0.9712)  loss_classifier: 0.4673 (0.4908)  loss_box_reg: 0.3838 (0.4038)  loss_objectness: 0.0489 (0.0584)  loss_rpn_box_reg: 0.0137 (0.0181)  time: 1.0652  data: 0.0198  max mem: 6508
Epoch: [4]  [100/104]  eta: 0:00:04  lr: 0.000100  loss: 0.9435 (0.9748)  loss_classifier: 0.4673 (0.4935)  loss_box_reg: 0.3838 (0.4057)  loss_objectness: 0.0452 (0.0574)  loss_rpn_box_reg: 0.0137 (0.0182)  time: 1.0699  data: 0.0194  max mem: 6508
Epoch: [4]  [103/104]  eta: 0:00:01  lr: 0.000100  loss: 0.9443 (0.9707)  loss_classifier: 0.4772 (0.4914)  loss_box_reg: 0.3838 (0.4038)  loss_objectness: 0.0440 (0.0574)  loss_rpn_box_reg: 0.0137 (0.0181)  time: 1.0685  data: 0.0189  max mem: 6508
Epoch: [4] Total time: 0:01:52 (1.0821 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:33  model_time: 0.5721 (0.5721)  evaluator_time: 0.0310 (0.0310)  time: 1.2730  data: 0.6565  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4422 (0.4454)  evaluator_time: 0.0264 (0.0278)  time: 0.4995  data: 0.0224  max mem: 6508
Test: Total time: 0:00:13 (0.5306 s / it)
Averaged stats: model_time: 0.4422 (0.4454)  evaluator_time: 0.0264 (0.0278)
Accumulating evaluation results...
DONE (t=0.20s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.027
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.071
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.013
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.035
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.027
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.004
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.012
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.084
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.119
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.140
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.151
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.025
Epoch: [5]  [  0/104]  eta: 0:03:18  lr: 0.000100  loss: 0.8089 (0.8089)  loss_classifier: 0.4197 (0.4197)  loss_box_reg: 0.3246 (0.3246)  loss_objectness: 0.0520 (0.0520)  loss_rpn_box_reg: 0.0127 (0.0127)  time: 1.9080  data: 0.7834  max mem: 6508
Epoch: [5]  [ 10/104]  eta: 0:01:50  lr: 0.000100  loss: 0.8925 (0.8742)  loss_classifier: 0.4646 (0.4502)  loss_box_reg: 0.3785 (0.3702)  loss_objectness: 0.0391 (0.0418)  loss_rpn_box_reg: 0.0127 (0.0121)  time: 1.1762  data: 0.0891  max mem: 6508
Epoch: [5]  [ 20/104]  eta: 0:01:36  lr: 0.000100  loss: 0.8562 (0.8700)  loss_classifier: 0.4325 (0.4480)  loss_box_reg: 0.3719 (0.3631)  loss_objectness: 0.0382 (0.0445)  loss_rpn_box_reg: 0.0129 (0.0144)  time: 1.1103  data: 0.0208  max mem: 6508
Epoch: [5]  [ 30/104]  eta: 0:01:23  lr: 0.000100  loss: 0.9047 (0.9220)  loss_classifier: 0.4699 (0.4718)  loss_box_reg: 0.3729 (0.3816)  loss_objectness: 0.0393 (0.0513)  loss_rpn_box_reg: 0.0143 (0.0173)  time: 1.1017  data: 0.0206  max mem: 6508
Epoch: [5]  [ 40/104]  eta: 0:01:11  lr: 0.000100  loss: 0.9314 (0.9486)  loss_classifier: 0.4938 (0.4789)  loss_box_reg: 0.3978 (0.3955)  loss_objectness: 0.0637 (0.0568)  loss_rpn_box_reg: 0.0147 (0.0175)  time: 1.0731  data: 0.0194  max mem: 6508
Epoch: [5]  [ 50/104]  eta: 0:00:59  lr: 0.000100  loss: 0.8906 (0.9427)  loss_classifier: 0.4345 (0.4750)  loss_box_reg: 0.3787 (0.3938)  loss_objectness: 0.0479 (0.0572)  loss_rpn_box_reg: 0.0133 (0.0167)  time: 1.0553  data: 0.0204  max mem: 6508
Epoch: [5]  [ 60/104]  eta: 0:00:48  lr: 0.000100  loss: 0.9511 (0.9705)  loss_classifier: 0.4656 (0.4899)  loss_box_reg: 0.3845 (0.4073)  loss_objectness: 0.0470 (0.0564)  loss_rpn_box_reg: 0.0150 (0.0169)  time: 1.0508  data: 0.0218  max mem: 6508
Epoch: [5]  [ 70/104]  eta: 0:00:36  lr: 0.000100  loss: 1.0254 (0.9822)  loss_classifier: 0.5239 (0.4963)  loss_box_reg: 0.4201 (0.4107)  loss_objectness: 0.0485 (0.0576)  loss_rpn_box_reg: 0.0179 (0.0175)  time: 1.0508  data: 0.0211  max mem: 6508
Epoch: [5]  [ 80/104]  eta: 0:00:25  lr: 0.000100  loss: 1.0060 (0.9836)  loss_classifier: 0.5202 (0.4971)  loss_box_reg: 0.4178 (0.4106)  loss_objectness: 0.0518 (0.0574)  loss_rpn_box_reg: 0.0204 (0.0185)  time: 1.0523  data: 0.0197  max mem: 6508
Epoch: [5]  [ 90/104]  eta: 0:00:15  lr: 0.000100  loss: 1.0086 (0.9848)  loss_classifier: 0.4859 (0.4972)  loss_box_reg: 0.4331 (0.4118)  loss_objectness: 0.0493 (0.0575)  loss_rpn_box_reg: 0.0192 (0.0184)  time: 1.0629  data: 0.0209  max mem: 6508
Epoch: [5]  [100/104]  eta: 0:00:04  lr: 0.000100  loss: 0.9006 (0.9680)  loss_classifier: 0.4366 (0.4888)  loss_box_reg: 0.3988 (0.4045)  loss_objectness: 0.0493 (0.0567)  loss_rpn_box_reg: 0.0140 (0.0180)  time: 1.0718  data: 0.0206  max mem: 6508
Epoch: [5]  [103/104]  eta: 0:00:01  lr: 0.000100  loss: 0.9371 (0.9711)  loss_classifier: 0.4366 (0.4910)  loss_box_reg: 0.4087 (0.4058)  loss_objectness: 0.0494 (0.0563)  loss_rpn_box_reg: 0.0170 (0.0180)  time: 1.0674  data: 0.0187  max mem: 6508
Epoch: [5] Total time: 0:01:52 (1.0808 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:33  model_time: 0.5164 (0.5164)  evaluator_time: 0.0355 (0.0355)  time: 1.3025  data: 0.7321  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4429 (0.4461)  evaluator_time: 0.0272 (0.0547)  time: 0.5069  data: 0.0215  max mem: 6508
Test: Total time: 0:00:14 (0.5703 s / it)
Averaged stats: model_time: 0.4429 (0.4461)  evaluator_time: 0.0272 (0.0547)
Accumulating evaluation results...
DONE (t=0.20s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.028
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.076
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.014
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.036
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.030
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.004
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.013
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.085
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.122
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.154
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.154
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.025
Epoch: [6]  [  0/104]  eta: 0:03:22  lr: 0.000010  loss: 1.2885 (1.2885)  loss_classifier: 0.6427 (0.6427)  loss_box_reg: 0.5884 (0.5884)  loss_objectness: 0.0387 (0.0387)  loss_rpn_box_reg: 0.0187 (0.0187)  time: 1.9494  data: 0.8234  max mem: 6508
Epoch: [6]  [ 10/104]  eta: 0:01:51  lr: 0.000010  loss: 1.0491 (1.0071)  loss_classifier: 0.5128 (0.5007)  loss_box_reg: 0.4600 (0.4501)  loss_objectness: 0.0419 (0.0397)  loss_rpn_box_reg: 0.0158 (0.0166)  time: 1.1824  data: 0.0937  max mem: 6508
Epoch: [6]  [ 20/104]  eta: 0:01:36  lr: 0.000010  loss: 0.9693 (0.9666)  loss_classifier: 0.4798 (0.4856)  loss_box_reg: 0.4007 (0.4134)  loss_objectness: 0.0451 (0.0497)  loss_rpn_box_reg: 0.0154 (0.0179)  time: 1.1121  data: 0.0219  max mem: 6508
Epoch: [6]  [ 30/104]  eta: 0:01:23  lr: 0.000010  loss: 0.9755 (0.9744)  loss_classifier: 0.4994 (0.4888)  loss_box_reg: 0.3828 (0.4143)  loss_objectness: 0.0528 (0.0530)  loss_rpn_box_reg: 0.0178 (0.0182)  time: 1.0983  data: 0.0216  max mem: 6508
Epoch: [6]  [ 40/104]  eta: 0:01:11  lr: 0.000010  loss: 0.9755 (0.9731)  loss_classifier: 0.4994 (0.4879)  loss_box_reg: 0.4114 (0.4137)  loss_objectness: 0.0538 (0.0540)  loss_rpn_box_reg: 0.0154 (0.0176)  time: 1.0684  data: 0.0194  max mem: 6508
Epoch: [6]  [ 50/104]  eta: 0:00:59  lr: 0.000010  loss: 0.9509 (0.9780)  loss_classifier: 0.4764 (0.4903)  loss_box_reg: 0.4237 (0.4153)  loss_objectness: 0.0538 (0.0546)  loss_rpn_box_reg: 0.0160 (0.0177)  time: 1.0579  data: 0.0201  max mem: 6508
Epoch: [6]  [ 60/104]  eta: 0:00:48  lr: 0.000010  loss: 1.0108 (0.9757)  loss_classifier: 0.5083 (0.4906)  loss_box_reg: 0.4370 (0.4141)  loss_objectness: 0.0433 (0.0535)  loss_rpn_box_reg: 0.0160 (0.0175)  time: 1.0574  data: 0.0230  max mem: 6508
Epoch: [6]  [ 70/104]  eta: 0:00:37  lr: 0.000010  loss: 0.8464 (0.9646)  loss_classifier: 0.4210 (0.4869)  loss_box_reg: 0.3655 (0.4074)  loss_objectness: 0.0396 (0.0526)  loss_rpn_box_reg: 0.0116 (0.0177)  time: 1.0579  data: 0.0235  max mem: 6508
Epoch: [6]  [ 80/104]  eta: 0:00:26  lr: 0.000010  loss: 0.8103 (0.9520)  loss_classifier: 0.4081 (0.4818)  loss_box_reg: 0.3511 (0.4010)  loss_objectness: 0.0350 (0.0518)  loss_rpn_box_reg: 0.0116 (0.0174)  time: 1.0585  data: 0.0215  max mem: 6508
Epoch: [6]  [ 90/104]  eta: 0:00:15  lr: 0.000010  loss: 0.9766 (0.9606)  loss_classifier: 0.4935 (0.4864)  loss_box_reg: 0.3786 (0.4059)  loss_objectness: 0.0320 (0.0510)  loss_rpn_box_reg: 0.0135 (0.0174)  time: 1.0676  data: 0.0202  max mem: 6508
Epoch: [6]  [100/104]  eta: 0:00:04  lr: 0.000010  loss: 1.0320 (0.9658)  loss_classifier: 0.4976 (0.4890)  loss_box_reg: 0.4053 (0.4066)  loss_objectness: 0.0450 (0.0526)  loss_rpn_box_reg: 0.0163 (0.0176)  time: 1.0749  data: 0.0204  max mem: 6508
Epoch: [6]  [103/104]  eta: 0:00:01  lr: 0.000010  loss: 1.0320 (0.9686)  loss_classifier: 0.4948 (0.4894)  loss_box_reg: 0.4036 (0.4065)  loss_objectness: 0.0528 (0.0548)  loss_rpn_box_reg: 0.0193 (0.0179)  time: 1.0718  data: 0.0201  max mem: 6508
Epoch: [6] Total time: 0:01:52 (1.0839 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:31  model_time: 0.5650 (0.5650)  evaluator_time: 0.0297 (0.0297)  time: 1.2216  data: 0.6035  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4417 (0.4446)  evaluator_time: 0.0195 (0.0298)  time: 0.4984  data: 0.0210  max mem: 6508
Test: Total time: 0:00:13 (0.5318 s / it)
Averaged stats: model_time: 0.4417 (0.4446)  evaluator_time: 0.0195 (0.0298)
Accumulating evaluation results...
DONE (t=0.20s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.028
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.076
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.013
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.036
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.029
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.004
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.013
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.085
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.122
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.155
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.153
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.025
Epoch: [7]  [  0/104]  eta: 0:03:27  lr: 0.000010  loss: 1.4218 (1.4218)  loss_classifier: 0.6696 (0.6696)  loss_box_reg: 0.6648 (0.6648)  loss_objectness: 0.0660 (0.0660)  loss_rpn_box_reg: 0.0214 (0.0214)  time: 1.9908  data: 0.8677  max mem: 6508
Epoch: [7]  [ 10/104]  eta: 0:01:50  lr: 0.000010  loss: 1.0095 (1.0092)  loss_classifier: 0.4840 (0.5016)  loss_box_reg: 0.4434 (0.4374)  loss_objectness: 0.0545 (0.0541)  loss_rpn_box_reg: 0.0150 (0.0161)  time: 1.1727  data: 0.0946  max mem: 6508
Epoch: [7]  [ 20/104]  eta: 0:01:36  lr: 0.000010  loss: 1.0162 (1.0178)  loss_classifier: 0.4840 (0.5060)  loss_box_reg: 0.4568 (0.4420)  loss_objectness: 0.0520 (0.0531)  loss_rpn_box_reg: 0.0150 (0.0168)  time: 1.1019  data: 0.0183  max mem: 6508
Epoch: [7]  [ 30/104]  eta: 0:01:23  lr: 0.000010  loss: 1.0405 (0.9977)  loss_classifier: 0.4935 (0.5015)  loss_box_reg: 0.3845 (0.4230)  loss_objectness: 0.0520 (0.0559)  loss_rpn_box_reg: 0.0181 (0.0173)  time: 1.1004  data: 0.0198  max mem: 6508
Epoch: [7]  [ 40/104]  eta: 0:01:11  lr: 0.000010  loss: 0.8868 (0.9853)  loss_classifier: 0.4536 (0.4965)  loss_box_reg: 0.3749 (0.4200)  loss_objectness: 0.0384 (0.0524)  loss_rpn_box_reg: 0.0131 (0.0164)  time: 1.0762  data: 0.0199  max mem: 6508
Epoch: [7]  [ 50/104]  eta: 0:00:59  lr: 0.000010  loss: 0.9059 (0.9954)  loss_classifier: 0.4536 (0.5012)  loss_box_reg: 0.3783 (0.4216)  loss_objectness: 0.0384 (0.0548)  loss_rpn_box_reg: 0.0150 (0.0178)  time: 1.0583  data: 0.0196  max mem: 6508
Epoch: [7]  [ 60/104]  eta: 0:00:48  lr: 0.000010  loss: 0.9683 (0.9907)  loss_classifier: 0.4741 (0.4998)  loss_box_reg: 0.4048 (0.4182)  loss_objectness: 0.0562 (0.0547)  loss_rpn_box_reg: 0.0206 (0.0180)  time: 1.0502  data: 0.0201  max mem: 6508
Epoch: [7]  [ 70/104]  eta: 0:00:36  lr: 0.000010  loss: 0.9683 (0.9950)  loss_classifier: 0.4857 (0.5028)  loss_box_reg: 0.4071 (0.4193)  loss_objectness: 0.0417 (0.0547)  loss_rpn_box_reg: 0.0165 (0.0182)  time: 1.0503  data: 0.0217  max mem: 6508
Epoch: [7]  [ 80/104]  eta: 0:00:25  lr: 0.000010  loss: 0.9465 (0.9916)  loss_classifier: 0.4834 (0.4989)  loss_box_reg: 0.4071 (0.4179)  loss_objectness: 0.0392 (0.0565)  loss_rpn_box_reg: 0.0139 (0.0183)  time: 1.0569  data: 0.0218  max mem: 6508
Epoch: [7]  [ 90/104]  eta: 0:00:15  lr: 0.000010  loss: 0.8826 (0.9701)  loss_classifier: 0.4435 (0.4892)  loss_box_reg: 0.3466 (0.4079)  loss_objectness: 0.0408 (0.0554)  loss_rpn_box_reg: 0.0126 (0.0176)  time: 1.0605  data: 0.0203  max mem: 6508
Epoch: [7]  [100/104]  eta: 0:00:04  lr: 0.000010  loss: 0.8826 (0.9735)  loss_classifier: 0.4553 (0.4914)  loss_box_reg: 0.3637 (0.4088)  loss_objectness: 0.0492 (0.0553)  loss_rpn_box_reg: 0.0123 (0.0181)  time: 1.0662  data: 0.0198  max mem: 6508
Epoch: [7]  [103/104]  eta: 0:00:01  lr: 0.000010  loss: 0.8776 (0.9691)  loss_classifier: 0.4504 (0.4894)  loss_box_reg: 0.3742 (0.4071)  loss_objectness: 0.0539 (0.0548)  loss_rpn_box_reg: 0.0123 (0.0179)  time: 1.0656  data: 0.0190  max mem: 6508
Epoch: [7] Total time: 0:01:52 (1.0809 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:51  model_time: 0.6856 (0.6856)  evaluator_time: 0.1232 (0.1232)  time: 1.9657  data: 1.1203  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4401 (0.4496)  evaluator_time: 0.0194 (0.0332)  time: 0.4916  data: 0.0202  max mem: 6508
Test: Total time: 0:00:14 (0.5636 s / it)
Averaged stats: model_time: 0.4401 (0.4496)  evaluator_time: 0.0194 (0.0332)
Accumulating evaluation results...
DONE (t=0.34s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.029
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.076
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.014
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.036
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.030
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.004
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.013
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.086
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.124
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.156
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.154
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.025
Epoch: [8]  [  0/104]  eta: 0:04:54  lr: 0.000010  loss: 1.4561 (1.4561)  loss_classifier: 0.6480 (0.6480)  loss_box_reg: 0.5642 (0.5642)  loss_objectness: 0.2178 (0.2178)  loss_rpn_box_reg: 0.0261 (0.0261)  time: 2.8293  data: 1.4711  max mem: 6508
Epoch: [8]  [ 10/104]  eta: 0:01:58  lr: 0.000010  loss: 1.0857 (1.0912)  loss_classifier: 0.5417 (0.5415)  loss_box_reg: 0.4551 (0.4548)  loss_objectness: 0.0692 (0.0732)  loss_rpn_box_reg: 0.0177 (0.0216)  time: 1.2569  data: 0.1488  max mem: 6508
Epoch: [8]  [ 20/104]  eta: 0:01:40  lr: 0.000010  loss: 1.0857 (1.0959)  loss_classifier: 0.5324 (0.5462)  loss_box_reg: 0.4345 (0.4633)  loss_objectness: 0.0358 (0.0657)  loss_rpn_box_reg: 0.0145 (0.0207)  time: 1.1151  data: 0.0215  max mem: 6508
Epoch: [8]  [ 30/104]  eta: 0:01:25  lr: 0.000010  loss: 0.9701 (1.0529)  loss_classifier: 0.5126 (0.5277)  loss_box_reg: 0.4303 (0.4479)  loss_objectness: 0.0384 (0.0583)  loss_rpn_box_reg: 0.0125 (0.0190)  time: 1.1088  data: 0.0248  max mem: 6508
Epoch: [8]  [ 40/104]  eta: 0:01:12  lr: 0.000010  loss: 0.9484 (1.0251)  loss_classifier: 0.4730 (0.5134)  loss_box_reg: 0.4157 (0.4351)  loss_objectness: 0.0396 (0.0582)  loss_rpn_box_reg: 0.0114 (0.0184)  time: 1.0783  data: 0.0226  max mem: 6508
Epoch: [8]  [ 50/104]  eta: 0:01:00  lr: 0.000010  loss: 0.8402 (0.9854)  loss_classifier: 0.4363 (0.4962)  loss_box_reg: 0.3572 (0.4167)  loss_objectness: 0.0386 (0.0549)  loss_rpn_box_reg: 0.0118 (0.0177)  time: 1.0613  data: 0.0218  max mem: 6508
Epoch: [8]  [ 60/104]  eta: 0:00:48  lr: 0.000010  loss: 0.8402 (0.9767)  loss_classifier: 0.4242 (0.4929)  loss_box_reg: 0.3395 (0.4116)  loss_objectness: 0.0426 (0.0539)  loss_rpn_box_reg: 0.0169 (0.0182)  time: 1.0469  data: 0.0207  max mem: 6508
Epoch: [8]  [ 70/104]  eta: 0:00:37  lr: 0.000010  loss: 0.9218 (0.9756)  loss_classifier: 0.4895 (0.4916)  loss_box_reg: 0.3836 (0.4118)  loss_objectness: 0.0431 (0.0543)  loss_rpn_box_reg: 0.0175 (0.0178)  time: 1.0490  data: 0.0219  max mem: 6508
Epoch: [8]  [ 80/104]  eta: 0:00:26  lr: 0.000010  loss: 0.9491 (0.9673)  loss_classifier: 0.4907 (0.4883)  loss_box_reg: 0.3861 (0.4068)  loss_objectness: 0.0524 (0.0548)  loss_rpn_box_reg: 0.0136 (0.0175)  time: 1.0606  data: 0.0241  max mem: 6508
Epoch: [8]  [ 90/104]  eta: 0:00:15  lr: 0.000010  loss: 0.9491 (0.9651)  loss_classifier: 0.4843 (0.4879)  loss_box_reg: 0.3844 (0.4062)  loss_objectness: 0.0402 (0.0535)  loss_rpn_box_reg: 0.0136 (0.0174)  time: 1.0663  data: 0.0230  max mem: 6508
Epoch: [8]  [100/104]  eta: 0:00:04  lr: 0.000010  loss: 0.9549 (0.9688)  loss_classifier: 0.4989 (0.4898)  loss_box_reg: 0.3844 (0.4074)  loss_objectness: 0.0385 (0.0541)  loss_rpn_box_reg: 0.0160 (0.0176)  time: 1.0727  data: 0.0209  max mem: 6508
Epoch: [8]  [103/104]  eta: 0:00:01  lr: 0.000010  loss: 0.9549 (0.9687)  loss_classifier: 0.4989 (0.4893)  loss_box_reg: 0.3844 (0.4069)  loss_objectness: 0.0552 (0.0546)  loss_rpn_box_reg: 0.0171 (0.0178)  time: 1.0754  data: 0.0204  max mem: 6508
Epoch: [8] Total time: 0:01:53 (1.0933 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:32  model_time: 0.5142 (0.5142)  evaluator_time: 0.0289 (0.0289)  time: 1.2555  data: 0.6940  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4425 (0.4429)  evaluator_time: 0.0257 (0.0398)  time: 0.4967  data: 0.0217  max mem: 6508
Test: Total time: 0:00:14 (0.5419 s / it)
Averaged stats: model_time: 0.4425 (0.4429)  evaluator_time: 0.0257 (0.0398)
Accumulating evaluation results...
DONE (t=0.20s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.029
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.077
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.014
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.037
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.030
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.004
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.014
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.086
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.125
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.157
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.155
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.025
Epoch: [9]  [  0/104]  eta: 0:03:31  lr: 0.000001  loss: 0.6304 (0.6304)  loss_classifier: 0.3522 (0.3522)  loss_box_reg: 0.2296 (0.2296)  loss_objectness: 0.0376 (0.0376)  loss_rpn_box_reg: 0.0110 (0.0110)  time: 2.0295  data: 0.9336  max mem: 6508
Epoch: [9]  [ 10/104]  eta: 0:01:51  lr: 0.000001  loss: 0.9999 (1.0055)  loss_classifier: 0.5041 (0.5084)  loss_box_reg: 0.4280 (0.4258)  loss_objectness: 0.0353 (0.0522)  loss_rpn_box_reg: 0.0127 (0.0191)  time: 1.1849  data: 0.1034  max mem: 6508
Epoch: [9]  [ 20/104]  eta: 0:01:36  lr: 0.000001  loss: 0.9999 (0.9711)  loss_classifier: 0.4866 (0.4879)  loss_box_reg: 0.4208 (0.4084)  loss_objectness: 0.0473 (0.0569)  loss_rpn_box_reg: 0.0161 (0.0179)  time: 1.1022  data: 0.0200  max mem: 6508
Epoch: [9]  [ 30/104]  eta: 0:01:23  lr: 0.000001  loss: 1.0741 (1.0154)  loss_classifier: 0.5166 (0.5073)  loss_box_reg: 0.4718 (0.4312)  loss_objectness: 0.0532 (0.0570)  loss_rpn_box_reg: 0.0161 (0.0199)  time: 1.0965  data: 0.0212  max mem: 6508
Epoch: [9]  [ 40/104]  eta: 0:01:11  lr: 0.000001  loss: 1.0208 (1.0040)  loss_classifier: 0.5166 (0.5043)  loss_box_reg: 0.4274 (0.4275)  loss_objectness: 0.0451 (0.0524)  loss_rpn_box_reg: 0.0175 (0.0197)  time: 1.0804  data: 0.0244  max mem: 6508
Epoch: [9]  [ 50/104]  eta: 0:00:59  lr: 0.000001  loss: 0.9630 (0.9806)  loss_classifier: 0.4896 (0.4954)  loss_box_reg: 0.4103 (0.4152)  loss_objectness: 0.0382 (0.0507)  loss_rpn_box_reg: 0.0177 (0.0192)  time: 1.0648  data: 0.0245  max mem: 6508
Epoch: [9]  [ 60/104]  eta: 0:00:48  lr: 0.000001  loss: 0.9851 (1.0015)  loss_classifier: 0.5003 (0.5055)  loss_box_reg: 0.4112 (0.4223)  loss_objectness: 0.0423 (0.0538)  loss_rpn_box_reg: 0.0188 (0.0200)  time: 1.0505  data: 0.0221  max mem: 6508
Epoch: [9]  [ 70/104]  eta: 0:00:36  lr: 0.000001  loss: 0.9349 (0.9778)  loss_classifier: 0.4848 (0.4952)  loss_box_reg: 0.3944 (0.4108)  loss_objectness: 0.0452 (0.0527)  loss_rpn_box_reg: 0.0173 (0.0191)  time: 1.0451  data: 0.0205  max mem: 6508
Epoch: [9]  [ 80/104]  eta: 0:00:26  lr: 0.000001  loss: 0.8820 (0.9653)  loss_classifier: 0.4339 (0.4890)  loss_box_reg: 0.3741 (0.4053)  loss_objectness: 0.0316 (0.0527)  loss_rpn_box_reg: 0.0098 (0.0183)  time: 1.0563  data: 0.0219  max mem: 6508
Epoch: [9]  [ 90/104]  eta: 0:00:15  lr: 0.000001  loss: 0.9438 (0.9706)  loss_classifier: 0.4703 (0.4897)  loss_box_reg: 0.3989 (0.4069)  loss_objectness: 0.0394 (0.0558)  loss_rpn_box_reg: 0.0126 (0.0182)  time: 1.0688  data: 0.0222  max mem: 6508
Epoch: [9]  [100/104]  eta: 0:00:04  lr: 0.000001  loss: 0.9698 (0.9680)  loss_classifier: 0.4863 (0.4885)  loss_box_reg: 0.4170 (0.4073)  loss_objectness: 0.0481 (0.0544)  loss_rpn_box_reg: 0.0124 (0.0177)  time: 1.0696  data: 0.0201  max mem: 6508
Epoch: [9]  [103/104]  eta: 0:00:01  lr: 0.000001  loss: 0.9841 (0.9693)  loss_classifier: 0.4922 (0.4894)  loss_box_reg: 0.4309 (0.4076)  loss_objectness: 0.0504 (0.0545)  loss_rpn_box_reg: 0.0142 (0.0178)  time: 1.0696  data: 0.0198  max mem: 6508
Epoch: [9] Total time: 0:01:52 (1.0823 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:37  model_time: 0.5051 (0.5051)  evaluator_time: 0.0168 (0.0168)  time: 1.4299  data: 0.8896  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4453 (0.4466)  evaluator_time: 0.0289 (0.0323)  time: 0.5133  data: 0.0253  max mem: 6508
Test: Total time: 0:00:14 (0.5489 s / it)
Averaged stats: model_time: 0.4453 (0.4466)  evaluator_time: 0.0289 (0.0323)
Accumulating evaluation results...
DONE (t=0.20s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.029
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.077
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.014
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.037
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.030
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.004
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.014
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.086
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.125
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.157
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.155
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.025
Epoch: [10]  [  0/104]  eta: 0:03:18  lr: 0.000001  loss: 0.8985 (0.8985)  loss_classifier: 0.4508 (0.4508)  loss_box_reg: 0.3753 (0.3753)  loss_objectness: 0.0546 (0.0546)  loss_rpn_box_reg: 0.0178 (0.0178)  time: 1.9127  data: 0.7313  max mem: 6508
Epoch: [10]  [ 10/104]  eta: 0:01:51  lr: 0.000001  loss: 0.9647 (0.9517)  loss_classifier: 0.4805 (0.4820)  loss_box_reg: 0.4047 (0.3959)  loss_objectness: 0.0524 (0.0555)  loss_rpn_box_reg: 0.0178 (0.0182)  time: 1.1822  data: 0.0876  max mem: 6508
Epoch: [10]  [ 20/104]  eta: 0:01:36  lr: 0.000001  loss: 0.8716 (0.9436)  loss_classifier: 0.4560 (0.4819)  loss_box_reg: 0.3817 (0.3931)  loss_objectness: 0.0483 (0.0518)  loss_rpn_box_reg: 0.0144 (0.0169)  time: 1.1104  data: 0.0225  max mem: 6508
Epoch: [10]  [ 30/104]  eta: 0:01:23  lr: 0.000001  loss: 0.8716 (0.9553)  loss_classifier: 0.4485 (0.4838)  loss_box_reg: 0.3739 (0.3985)  loss_objectness: 0.0557 (0.0558)  loss_rpn_box_reg: 0.0143 (0.0172)  time: 1.0981  data: 0.0211  max mem: 6508
Epoch: [10]  [ 40/104]  eta: 0:01:11  lr: 0.000001  loss: 0.9804 (0.9572)  loss_classifier: 0.4722 (0.4811)  loss_box_reg: 0.4234 (0.3979)  loss_objectness: 0.0491 (0.0607)  loss_rpn_box_reg: 0.0168 (0.0175)  time: 1.0732  data: 0.0198  max mem: 6508
Epoch: [10]  [ 50/104]  eta: 0:00:59  lr: 0.000001  loss: 0.9462 (0.9525)  loss_classifier: 0.4719 (0.4801)  loss_box_reg: 0.3912 (0.3961)  loss_objectness: 0.0443 (0.0591)  loss_rpn_box_reg: 0.0157 (0.0172)  time: 1.0611  data: 0.0215  max mem: 6508
Epoch: [10]  [ 60/104]  eta: 0:00:48  lr: 0.000001  loss: 0.9680 (0.9716)  loss_classifier: 0.5065 (0.4896)  loss_box_reg: 0.4100 (0.4073)  loss_objectness: 0.0416 (0.0573)  loss_rpn_box_reg: 0.0134 (0.0175)  time: 1.0651  data: 0.0244  max mem: 6508
Epoch: [10]  [ 70/104]  eta: 0:00:37  lr: 0.000001  loss: 0.9544 (0.9686)  loss_classifier: 0.4950 (0.4873)  loss_box_reg: 0.4145 (0.4071)  loss_objectness: 0.0416 (0.0567)  loss_rpn_box_reg: 0.0155 (0.0175)  time: 1.0577  data: 0.0222  max mem: 6508
Epoch: [10]  [ 80/104]  eta: 0:00:26  lr: 0.000001  loss: 0.9351 (0.9720)  loss_classifier: 0.4696 (0.4891)  loss_box_reg: 0.3985 (0.4081)  loss_objectness: 0.0418 (0.0572)  loss_rpn_box_reg: 0.0155 (0.0177)  time: 1.0524  data: 0.0197  max mem: 6508
Epoch: [10]  [ 90/104]  eta: 0:00:15  lr: 0.000001  loss: 0.9478 (0.9635)  loss_classifier: 0.4696 (0.4860)  loss_box_reg: 0.3947 (0.4043)  loss_objectness: 0.0492 (0.0556)  loss_rpn_box_reg: 0.0123 (0.0176)  time: 1.0638  data: 0.0208  max mem: 6508
Epoch: [10]  [100/104]  eta: 0:00:04  lr: 0.000001  loss: 0.9310 (0.9661)  loss_classifier: 0.4724 (0.4876)  loss_box_reg: 0.4140 (0.4064)  loss_objectness: 0.0444 (0.0546)  loss_rpn_box_reg: 0.0123 (0.0175)  time: 1.0693  data: 0.0202  max mem: 6508
Epoch: [10]  [103/104]  eta: 0:00:01  lr: 0.000001  loss: 0.9491 (0.9697)  loss_classifier: 0.4825 (0.4894)  loss_box_reg: 0.4180 (0.4079)  loss_objectness: 0.0472 (0.0545)  loss_rpn_box_reg: 0.0167 (0.0178)  time: 1.0690  data: 0.0190  max mem: 6508
Epoch: [10] Total time: 0:01:52 (1.0835 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:32  model_time: 0.5264 (0.5264)  evaluator_time: 0.0305 (0.0305)  time: 1.2548  data: 0.6806  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4470 (0.4466)  evaluator_time: 0.0282 (0.0290)  time: 0.5045  data: 0.0221  max mem: 6508
Test: Total time: 0:00:13 (0.5355 s / it)
Averaged stats: model_time: 0.4470 (0.4466)  evaluator_time: 0.0282 (0.0290)
Accumulating evaluation results...
DONE (t=0.22s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.028
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.076
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.014
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.036
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.030
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.004
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.014
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.086
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.125
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.150
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.155
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.025
Epoch: [11]  [  0/104]  eta: 0:03:29  lr: 0.000001  loss: 0.7550 (0.7550)  loss_classifier: 0.4027 (0.4027)  loss_box_reg: 0.3002 (0.3002)  loss_objectness: 0.0430 (0.0430)  loss_rpn_box_reg: 0.0090 (0.0090)  time: 2.0159  data: 0.9170  max mem: 6508
Epoch: [11]  [ 10/104]  eta: 0:01:51  lr: 0.000001  loss: 0.9419 (0.9750)  loss_classifier: 0.4774 (0.4959)  loss_box_reg: 0.4060 (0.4132)  loss_objectness: 0.0463 (0.0481)  loss_rpn_box_reg: 0.0138 (0.0178)  time: 1.1869  data: 0.1023  max mem: 6508
Epoch: [11]  [ 20/104]  eta: 0:01:36  lr: 0.000001  loss: 1.0021 (1.0197)  loss_classifier: 0.5084 (0.5174)  loss_box_reg: 0.4273 (0.4354)  loss_objectness: 0.0445 (0.0490)  loss_rpn_box_reg: 0.0159 (0.0179)  time: 1.1058  data: 0.0228  max mem: 6508
Epoch: [11]  [ 30/104]  eta: 0:01:23  lr: 0.000001  loss: 0.9841 (0.9790)  loss_classifier: 0.5064 (0.5001)  loss_box_reg: 0.4191 (0.4162)  loss_objectness: 0.0389 (0.0471)  loss_rpn_box_reg: 0.0142 (0.0157)  time: 1.0936  data: 0.0238  max mem: 6508
Epoch: [11]  [ 40/104]  eta: 0:01:11  lr: 0.000001  loss: 0.8633 (0.9623)  loss_classifier: 0.4275 (0.4892)  loss_box_reg: 0.3595 (0.4084)  loss_objectness: 0.0427 (0.0494)  loss_rpn_box_reg: 0.0124 (0.0153)  time: 1.0685  data: 0.0212  max mem: 6508
Epoch: [11]  [ 50/104]  eta: 0:00:59  lr: 0.000001  loss: 0.9035 (0.9643)  loss_classifier: 0.4459 (0.4886)  loss_box_reg: 0.3850 (0.4086)  loss_objectness: 0.0444 (0.0513)  loss_rpn_box_reg: 0.0146 (0.0158)  time: 1.0535  data: 0.0200  max mem: 6508
Epoch: [11]  [ 60/104]  eta: 0:00:48  lr: 0.000001  loss: 0.8914 (0.9436)  loss_classifier: 0.4438 (0.4784)  loss_box_reg: 0.3755 (0.3993)  loss_objectness: 0.0444 (0.0507)  loss_rpn_box_reg: 0.0118 (0.0153)  time: 1.0539  data: 0.0221  max mem: 6508
Epoch: [11]  [ 70/104]  eta: 0:00:36  lr: 0.000001  loss: 0.8592 (0.9572)  loss_classifier: 0.4411 (0.4834)  loss_box_reg: 0.3648 (0.4021)  loss_objectness: 0.0479 (0.0546)  loss_rpn_box_reg: 0.0142 (0.0170)  time: 1.0521  data: 0.0216  max mem: 6508
Epoch: [11]  [ 80/104]  eta: 0:00:25  lr: 0.000001  loss: 0.8975 (0.9558)  loss_classifier: 0.4754 (0.4825)  loss_box_reg: 0.3745 (0.4016)  loss_objectness: 0.0491 (0.0543)  loss_rpn_box_reg: 0.0185 (0.0174)  time: 1.0519  data: 0.0195  max mem: 6508
Epoch: [11]  [ 90/104]  eta: 0:00:15  lr: 0.000001  loss: 0.8975 (0.9679)  loss_classifier: 0.4754 (0.4883)  loss_box_reg: 0.3767 (0.4061)  loss_objectness: 0.0581 (0.0557)  loss_rpn_box_reg: 0.0180 (0.0178)  time: 1.0667  data: 0.0204  max mem: 6508
Epoch: [11]  [100/104]  eta: 0:00:04  lr: 0.000001  loss: 0.9923 (0.9718)  loss_classifier: 0.4808 (0.4895)  loss_box_reg: 0.3905 (0.4083)  loss_objectness: 0.0560 (0.0561)  loss_rpn_box_reg: 0.0173 (0.0179)  time: 1.0735  data: 0.0206  max mem: 6508
Epoch: [11]  [103/104]  eta: 0:00:01  lr: 0.000001  loss: 0.8953 (0.9701)  loss_classifier: 0.4682 (0.4893)  loss_box_reg: 0.3905 (0.4076)  loss_objectness: 0.0488 (0.0553)  loss_rpn_box_reg: 0.0153 (0.0178)  time: 1.0683  data: 0.0192  max mem: 6508
Epoch: [11] Total time: 0:01:52 (1.0808 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:32  model_time: 0.4939 (0.4939)  evaluator_time: 0.0360 (0.0360)  time: 1.2434  data: 0.6966  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4436 (0.4446)  evaluator_time: 0.0245 (0.0287)  time: 0.5009  data: 0.0231  max mem: 6508
Test: Total time: 0:00:14 (0.5470 s / it)
Averaged stats: model_time: 0.4436 (0.4446)  evaluator_time: 0.0245 (0.0287)
Accumulating evaluation results...
DONE (t=0.20s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.028
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.076
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.014
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.036
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.030
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.004
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.014
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.086
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.125
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.150
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.155
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.025
Epoch: [12]  [  0/104]  eta: 0:03:17  lr: 0.000000  loss: 1.2637 (1.2637)  loss_classifier: 0.5983 (0.5983)  loss_box_reg: 0.6241 (0.6241)  loss_objectness: 0.0310 (0.0310)  loss_rpn_box_reg: 0.0103 (0.0103)  time: 1.9002  data: 0.7437  max mem: 6508
Epoch: [12]  [ 10/104]  eta: 0:01:50  lr: 0.000000  loss: 1.1292 (1.1048)  loss_classifier: 0.5538 (0.5505)  loss_box_reg: 0.4678 (0.4885)  loss_objectness: 0.0361 (0.0473)  loss_rpn_box_reg: 0.0126 (0.0185)  time: 1.1738  data: 0.0860  max mem: 6508
Epoch: [12]  [ 20/104]  eta: 0:01:35  lr: 0.000000  loss: 0.9303 (1.0023)  loss_classifier: 0.4809 (0.5046)  loss_box_reg: 0.4035 (0.4351)  loss_objectness: 0.0362 (0.0456)  loss_rpn_box_reg: 0.0159 (0.0171)  time: 1.1010  data: 0.0209  max mem: 6508
Epoch: [12]  [ 30/104]  eta: 0:01:22  lr: 0.000000  loss: 0.8486 (0.9891)  loss_classifier: 0.4486 (0.4980)  loss_box_reg: 0.3564 (0.4227)  loss_objectness: 0.0408 (0.0495)  loss_rpn_box_reg: 0.0159 (0.0189)  time: 1.0926  data: 0.0215  max mem: 6508
Epoch: [12]  [ 40/104]  eta: 0:01:10  lr: 0.000000  loss: 0.8770 (0.9793)  loss_classifier: 0.4563 (0.4950)  loss_box_reg: 0.3876 (0.4167)  loss_objectness: 0.0418 (0.0494)  loss_rpn_box_reg: 0.0150 (0.0182)  time: 1.0717  data: 0.0202  max mem: 6508
Epoch: [12]  [ 50/104]  eta: 0:00:59  lr: 0.000000  loss: 0.9349 (0.9718)  loss_classifier: 0.4563 (0.4900)  loss_box_reg: 0.3876 (0.4106)  loss_objectness: 0.0422 (0.0525)  loss_rpn_box_reg: 0.0155 (0.0187)  time: 1.0530  data: 0.0187  max mem: 6508
Epoch: [12]  [ 60/104]  eta: 0:00:47  lr: 0.000000  loss: 0.8260 (0.9482)  loss_classifier: 0.4230 (0.4795)  loss_box_reg: 0.3398 (0.4001)  loss_objectness: 0.0408 (0.0506)  loss_rpn_box_reg: 0.0120 (0.0181)  time: 1.0511  data: 0.0196  max mem: 6508
Epoch: [12]  [ 70/104]  eta: 0:00:36  lr: 0.000000  loss: 0.8678 (0.9552)  loss_classifier: 0.4651 (0.4838)  loss_box_reg: 0.3602 (0.4023)  loss_objectness: 0.0410 (0.0511)  loss_rpn_box_reg: 0.0118 (0.0180)  time: 1.0573  data: 0.0204  max mem: 6508
Epoch: [12]  [ 80/104]  eta: 0:00:25  lr: 0.000000  loss: 0.9021 (0.9517)  loss_classifier: 0.4667 (0.4829)  loss_box_reg: 0.3757 (0.4006)  loss_objectness: 0.0441 (0.0507)  loss_rpn_box_reg: 0.0143 (0.0176)  time: 1.0608  data: 0.0206  max mem: 6508
Epoch: [12]  [ 90/104]  eta: 0:00:15  lr: 0.000000  loss: 0.9521 (0.9610)  loss_classifier: 0.4933 (0.4853)  loss_box_reg: 0.3953 (0.4027)  loss_objectness: 0.0518 (0.0549)  loss_rpn_box_reg: 0.0153 (0.0181)  time: 1.0658  data: 0.0224  max mem: 6508
Epoch: [12]  [100/104]  eta: 0:00:04  lr: 0.000000  loss: 0.9724 (0.9599)  loss_classifier: 0.4933 (0.4856)  loss_box_reg: 0.4017 (0.4031)  loss_objectness: 0.0518 (0.0537)  loss_rpn_box_reg: 0.0148 (0.0176)  time: 1.0764  data: 0.0239  max mem: 6508
Epoch: [12]  [103/104]  eta: 0:00:01  lr: 0.000000  loss: 0.9875 (0.9686)  loss_classifier: 0.5226 (0.4893)  loss_box_reg: 0.4037 (0.4070)  loss_objectness: 0.0518 (0.0545)  loss_rpn_box_reg: 0.0186 (0.0178)  time: 1.0762  data: 0.0228  max mem: 6508
Epoch: [12] Total time: 0:01:52 (1.0816 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:32  model_time: 0.5460 (0.5460)  evaluator_time: 0.0506 (0.0506)  time: 1.2540  data: 0.6371  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4400 (0.4437)  evaluator_time: 0.0189 (0.0318)  time: 0.4944  data: 0.0212  max mem: 6508
Test: Total time: 0:00:13 (0.5340 s / it)
Averaged stats: model_time: 0.4400 (0.4437)  evaluator_time: 0.0189 (0.0318)
Accumulating evaluation results...
DONE (t=0.21s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.028
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.076
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.014
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.036
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.030
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.004
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.014
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.086
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.125
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.150
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.155
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.025
Epoch: [13]  [  0/104]  eta: 0:03:15  lr: 0.000000  loss: 1.0455 (1.0455)  loss_classifier: 0.5254 (0.5254)  loss_box_reg: 0.4590 (0.4590)  loss_objectness: 0.0490 (0.0490)  loss_rpn_box_reg: 0.0121 (0.0121)  time: 1.8814  data: 0.6603  max mem: 6508
Epoch: [13]  [ 10/104]  eta: 0:01:50  lr: 0.000000  loss: 0.9119 (1.0463)  loss_classifier: 0.4604 (0.5150)  loss_box_reg: 0.3810 (0.4381)  loss_objectness: 0.0483 (0.0701)  loss_rpn_box_reg: 0.0173 (0.0231)  time: 1.1755  data: 0.0781  max mem: 6508
Epoch: [13]  [ 20/104]  eta: 0:01:35  lr: 0.000000  loss: 1.1687 (1.1211)  loss_classifier: 0.5784 (0.5539)  loss_box_reg: 0.5096 (0.4756)  loss_objectness: 0.0587 (0.0683)  loss_rpn_box_reg: 0.0221 (0.0233)  time: 1.1049  data: 0.0194  max mem: 6508
Epoch: [13]  [ 30/104]  eta: 0:01:23  lr: 0.000000  loss: 1.0182 (1.0474)  loss_classifier: 0.4869 (0.5202)  loss_box_reg: 0.4253 (0.4426)  loss_objectness: 0.0570 (0.0642)  loss_rpn_box_reg: 0.0168 (0.0204)  time: 1.0998  data: 0.0205  max mem: 6508
Epoch: [13]  [ 40/104]  eta: 0:01:11  lr: 0.000000  loss: 0.8346 (1.0101)  loss_classifier: 0.4342 (0.5053)  loss_box_reg: 0.3361 (0.4237)  loss_objectness: 0.0438 (0.0615)  loss_rpn_box_reg: 0.0110 (0.0196)  time: 1.0775  data: 0.0208  max mem: 6508
Epoch: [13]  [ 50/104]  eta: 0:00:59  lr: 0.000000  loss: 0.8370 (0.9929)  loss_classifier: 0.4228 (0.4970)  loss_box_reg: 0.3397 (0.4177)  loss_objectness: 0.0368 (0.0584)  loss_rpn_box_reg: 0.0137 (0.0198)  time: 1.0545  data: 0.0194  max mem: 6508
Epoch: [13]  [ 60/104]  eta: 0:00:47  lr: 0.000000  loss: 0.8682 (0.9842)  loss_classifier: 0.4398 (0.4952)  loss_box_reg: 0.3525 (0.4129)  loss_objectness: 0.0405 (0.0568)  loss_rpn_box_reg: 0.0125 (0.0194)  time: 1.0480  data: 0.0208  max mem: 6508
Epoch: [13]  [ 70/104]  eta: 0:00:36  lr: 0.000000  loss: 0.9294 (0.9796)  loss_classifier: 0.4642 (0.4935)  loss_box_reg: 0.3764 (0.4101)  loss_objectness: 0.0451 (0.0570)  loss_rpn_box_reg: 0.0128 (0.0189)  time: 1.0484  data: 0.0222  max mem: 6508
Epoch: [13]  [ 80/104]  eta: 0:00:25  lr: 0.000000  loss: 0.9831 (0.9836)  loss_classifier: 0.4986 (0.4969)  loss_box_reg: 0.4131 (0.4129)  loss_objectness: 0.0478 (0.0551)  loss_rpn_box_reg: 0.0153 (0.0187)  time: 1.0544  data: 0.0215  max mem: 6508
Epoch: [13]  [ 90/104]  eta: 0:00:15  lr: 0.000000  loss: 0.9340 (0.9727)  loss_classifier: 0.4654 (0.4910)  loss_box_reg: 0.4099 (0.4078)  loss_objectness: 0.0500 (0.0554)  loss_rpn_box_reg: 0.0152 (0.0185)  time: 1.0612  data: 0.0203  max mem: 6508
Epoch: [13]  [100/104]  eta: 0:00:04  lr: 0.000000  loss: 0.9029 (0.9735)  loss_classifier: 0.4607 (0.4907)  loss_box_reg: 0.3724 (0.4095)  loss_objectness: 0.0532 (0.0552)  loss_rpn_box_reg: 0.0149 (0.0181)  time: 1.0731  data: 0.0209  max mem: 6508
Epoch: [13]  [103/104]  eta: 0:00:01  lr: 0.000000  loss: 0.9214 (0.9691)  loss_classifier: 0.4654 (0.4893)  loss_box_reg: 0.3794 (0.4073)  loss_objectness: 0.0520 (0.0547)  loss_rpn_box_reg: 0.0129 (0.0178)  time: 1.0757  data: 0.0207  max mem: 6508
Epoch: [13] Total time: 0:01:52 (1.0814 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:50  model_time: 0.6735 (0.6735)  evaluator_time: 0.1359 (0.1359)  time: 1.9361  data: 1.0955  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4414 (0.4503)  evaluator_time: 0.0201 (0.0303)  time: 0.4906  data: 0.0192  max mem: 6508
Test: Total time: 0:00:14 (0.5563 s / it)
Averaged stats: model_time: 0.4414 (0.4503)  evaluator_time: 0.0201 (0.0303)
Accumulating evaluation results...
DONE (t=0.35s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.028
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.076
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.014
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.036
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.030
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.004
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.014
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.086
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.125
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.150
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.155
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.025
Epoch: [14]  [  0/104]  eta: 0:04:44  lr: 0.000000  loss: 0.6101 (0.6101)  loss_classifier: 0.2887 (0.2887)  loss_box_reg: 0.2060 (0.2060)  loss_objectness: 0.0904 (0.0904)  loss_rpn_box_reg: 0.0251 (0.0251)  time: 2.7340  data: 1.3935  max mem: 6508
Epoch: [14]  [ 10/104]  eta: 0:01:55  lr: 0.000000  loss: 0.9464 (0.9770)  loss_classifier: 0.4903 (0.4883)  loss_box_reg: 0.3871 (0.4116)  loss_objectness: 0.0569 (0.0572)  loss_rpn_box_reg: 0.0206 (0.0199)  time: 1.2336  data: 0.1411  max mem: 6508
Epoch: [14]  [ 20/104]  eta: 0:01:38  lr: 0.000000  loss: 0.9644 (0.9994)  loss_classifier: 0.4903 (0.4978)  loss_box_reg: 0.4017 (0.4186)  loss_objectness: 0.0541 (0.0616)  loss_rpn_box_reg: 0.0206 (0.0214)  time: 1.0910  data: 0.0204  max mem: 6508
Epoch: [14]  [ 30/104]  eta: 0:01:24  lr: 0.000000  loss: 0.9311 (0.9635)  loss_classifier: 0.4634 (0.4811)  loss_box_reg: 0.3893 (0.4005)  loss_objectness: 0.0541 (0.0622)  loss_rpn_box_reg: 0.0179 (0.0197)  time: 1.0857  data: 0.0232  max mem: 6508
Epoch: [14]  [ 40/104]  eta: 0:01:11  lr: 0.000000  loss: 0.9757 (0.9812)  loss_classifier: 0.4914 (0.4918)  loss_box_reg: 0.4169 (0.4124)  loss_objectness: 0.0434 (0.0582)  loss_rpn_box_reg: 0.0167 (0.0188)  time: 1.0682  data: 0.0206  max mem: 6508
Epoch: [14]  [ 50/104]  eta: 0:00:59  lr: 0.000000  loss: 1.0226 (0.9578)  loss_classifier: 0.5268 (0.4829)  loss_box_reg: 0.4330 (0.4021)  loss_objectness: 0.0329 (0.0543)  loss_rpn_box_reg: 0.0153 (0.0185)  time: 1.0619  data: 0.0200  max mem: 6508
Epoch: [14]  [ 60/104]  eta: 0:00:48  lr: 0.000000  loss: 0.8524 (0.9526)  loss_classifier: 0.4663 (0.4810)  loss_box_reg: 0.3546 (0.4005)  loss_objectness: 0.0433 (0.0537)  loss_rpn_box_reg: 0.0134 (0.0175)  time: 1.0578  data: 0.0199  max mem: 6508
Epoch: [14]  [ 70/104]  eta: 0:00:37  lr: 0.000000  loss: 0.9881 (0.9633)  loss_classifier: 0.4924 (0.4869)  loss_box_reg: 0.3968 (0.4038)  loss_objectness: 0.0472 (0.0547)  loss_rpn_box_reg: 0.0161 (0.0180)  time: 1.0619  data: 0.0208  max mem: 6508
Epoch: [14]  [ 80/104]  eta: 0:00:26  lr: 0.000000  loss: 0.9881 (0.9635)  loss_classifier: 0.5019 (0.4880)  loss_box_reg: 0.3968 (0.4039)  loss_objectness: 0.0436 (0.0536)  loss_rpn_box_reg: 0.0161 (0.0180)  time: 1.0679  data: 0.0220  max mem: 6508
Epoch: [14]  [ 90/104]  eta: 0:00:15  lr: 0.000000  loss: 1.0380 (0.9768)  loss_classifier: 0.5328 (0.4938)  loss_box_reg: 0.4397 (0.4100)  loss_objectness: 0.0436 (0.0547)  loss_rpn_box_reg: 0.0171 (0.0183)  time: 1.0679  data: 0.0217  max mem: 6508
Epoch: [14]  [100/104]  eta: 0:00:04  lr: 0.000000  loss: 0.9153 (0.9713)  loss_classifier: 0.4701 (0.4910)  loss_box_reg: 0.3913 (0.4084)  loss_objectness: 0.0432 (0.0538)  loss_rpn_box_reg: 0.0161 (0.0181)  time: 1.0648  data: 0.0199  max mem: 6508
Epoch: [14]  [103/104]  eta: 0:00:01  lr: 0.000000  loss: 0.9153 (0.9683)  loss_classifier: 0.4701 (0.4895)  loss_box_reg: 0.3900 (0.4075)  loss_objectness: 0.0465 (0.0535)  loss_rpn_box_reg: 0.0154 (0.0179)  time: 1.0654  data: 0.0198  max mem: 6508
Epoch: [14] Total time: 0:01:53 (1.0872 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:34  model_time: 0.5303 (0.5303)  evaluator_time: 0.0218 (0.0218)  time: 1.3316  data: 0.7626  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4412 (0.4442)  evaluator_time: 0.0285 (0.0312)  time: 0.5028  data: 0.0216  max mem: 6508
Test: Total time: 0:00:13 (0.5376 s / it)
Averaged stats: model_time: 0.4412 (0.4442)  evaluator_time: 0.0285 (0.0312)
Accumulating evaluation results...
DONE (t=0.19s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.028
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.076
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.014
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.036
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.030
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.004
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.014
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.086
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.125
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.150
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.155
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.025

#save adelta
import pickle
Filename = "FRCNN4adelta.pkl"
# Define the file path where you want to save the model
filename = "/content/drive/MyDrive/dataset1/FRCNN4adelta.pkl"

# Save the model to the specified file path
torch.save(model.state_dict(), filename)
# Save the Modle to file in the current working directory
with open(Filename, 'wb') as file:
    pickle.dump(model, file)
# Load the Model back from file
with open(Filename, 'rb') as file:
    model = pickle.load(file)
model

FasterRCNN(
  (transform): GeneralizedRCNNTransform(
      Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
      Resize(min_size=(800,), max_size=1333, mode='bilinear')
  )
  (backbone): BackboneWithFPN(
    (body): IntermediateLayerGetter(
      (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)
      (bn1): FrozenBatchNorm2d(64, eps=0.0)
      (relu): ReLU(inplace=True)
      (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)
      (layer1): Sequential(
        (0): Bottleneck(
          (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(64, eps=0.0)
          (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(64, eps=0.0)
          (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(256, eps=0.0)
          (relu): ReLU(inplace=True)
          (downsample): Sequential(
            (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
            (1): FrozenBatchNorm2d(256, eps=0.0)
          )
        )
        (1): Bottleneck(
          (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(64, eps=0.0)
          (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(64, eps=0.0)
          (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(256, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (2): Bottleneck(
          (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(64, eps=0.0)
          (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(64, eps=0.0)
          (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(256, eps=0.0)
          (relu): ReLU(inplace=True)
        )
      )
      (layer2): Sequential(
        (0): Bottleneck(
          (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(128, eps=0.0)
          (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(128, eps=0.0)
          (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(512, eps=0.0)
          (relu): ReLU(inplace=True)
          (downsample): Sequential(
            (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)
            (1): FrozenBatchNorm2d(512, eps=0.0)
          )
        )
        (1): Bottleneck(
          (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(128, eps=0.0)
          (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(128, eps=0.0)
          (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(512, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (2): Bottleneck(
          (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(128, eps=0.0)
          (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(128, eps=0.0)
          (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(512, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (3): Bottleneck(
          (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(128, eps=0.0)
          (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(128, eps=0.0)
          (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(512, eps=0.0)
          (relu): ReLU(inplace=True)
        )
      )
      (layer3): Sequential(
        (0): Bottleneck(
          (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
          (downsample): Sequential(
            (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False)
            (1): FrozenBatchNorm2d(1024, eps=0.0)
          )
        )
        (1): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (2): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (3): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (4): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (5): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
        )
      )
      (layer4): Sequential(
        (0): Bottleneck(
          (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(512, eps=0.0)
          (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(512, eps=0.0)
          (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(2048, eps=0.0)
          (relu): ReLU(inplace=True)
          (downsample): Sequential(
            (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False)
            (1): FrozenBatchNorm2d(2048, eps=0.0)
          )
        )
        (1): Bottleneck(
          (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(512, eps=0.0)
          (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(512, eps=0.0)
          (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(2048, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (2): Bottleneck(
          (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(512, eps=0.0)
          (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(512, eps=0.0)
          (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(2048, eps=0.0)
          (relu): ReLU(inplace=True)
        )
      )
    )
    (fpn): FeaturePyramidNetwork(
      (inner_blocks): ModuleList(
        (0): Conv2dNormActivation(
          (0): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1))
        )
        (1): Conv2dNormActivation(
          (0): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1))
        )
        (2): Conv2dNormActivation(
          (0): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1))
        )
        (3): Conv2dNormActivation(
          (0): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1))
        )
      )
      (layer_blocks): ModuleList(
        (0-3): 4 x Conv2dNormActivation(
          (0): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
        )
      )
      (extra_blocks): LastLevelMaxPool()
    )
  )
  (rpn): RegionProposalNetwork(
    (anchor_generator): AnchorGenerator()
    (head): RPNHead(
      (conv): Sequential(
        (0): Conv2dNormActivation(
          (0): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
          (1): ReLU(inplace=True)
        )
      )
      (cls_logits): Conv2d(256, 3, kernel_size=(1, 1), stride=(1, 1))
      (bbox_pred): Conv2d(256, 12, kernel_size=(1, 1), stride=(1, 1))
    )
  )
  (roi_heads): RoIHeads(
    (box_roi_pool): MultiScaleRoIAlign(featmap_names=['0', '1', '2', '3'], output_size=(7, 7), sampling_ratio=2)
    (box_head): TwoMLPHead(
      (fc6): Linear(in_features=12544, out_features=1024, bias=True)
      (fc7): Linear(in_features=1024, out_features=1024, bias=True)
    )
    (box_predictor): FastRCNNPredictor(
      (cls_score): Linear(in_features=1024, out_features=11, bias=True)
      (bbox_pred): Linear(in_features=1024, out_features=44, bias=True)
    )
  )
)

#adam_sgd
import torch
import torch.optim as optim
from torch.optim.lr_scheduler import StepLR

def hybrid_optimizer(model, params_adam, params_sgd, lr=0.001, momentum=0.9, weight_decay=0.0005, step_size=3, gamma=0.1):
    # Initialize Adam optimizer for initial training
    optimizer_adam = optim.Adam(params_adam, lr=lr, weight_decay=weight_decay)

    # Initialize SGD optimizer for switching
    optimizer_sgd = optim.SGD(params_sgd, lr=lr, momentum=momentum, weight_decay=weight_decay)

    # Learning rate scheduler for SGD optimizer
    lr_scheduler = StepLR(optimizer_sgd, step_size=step_size, gamma=gamma)

    # Initial optimizer
    optimizer = optimizer_adam

    return optimizer, optimizer_adam, optimizer_sgd, lr_scheduler

# Define your model
num_classes = 11
model = get_object_detection_model(num_classes)

# Move model to the right device
device = torch.device('cuda') if torch.cuda.is_available() else torch.device('cpu')
model.to(device)

# Get model parameters for Adam and SGD
params_adam = [p for p in model.parameters() if p.requires_grad]
params_sgd = [p for p in model.parameters() if p.requires_grad]

# Initialize hybrid optimizer
optimizer, optimizer_adam, optimizer_sgd, lr_scheduler = hybrid_optimizer(model, params_adam, params_sgd)

# Training loop
num_epochs = 15
hybrid_results = []  # Initialize list to store hybrid technique values
is_adam = True  # Flag to indicate whether the current optimizer is Adam or not

for epoch in range(num_epochs):
    # Train using current optimizer
    train_one_epoch(model, optimizer, data_loader, device, epoch, print_freq=10)

    # Append value of hybrid technique
    hybrid_results.append(int(is_adam))  # Append 1 for Adam, 0 for SGD

    # Check a condition to switch to SGD
    if epoch == 5:  # For example, switch to SGD after 5 epochs
        optimizer = optimizer_sgd
        lr_scheduler.step()  # Update learning rate for SGD
        is_adam = False

    # Evaluate on the test dataset
    evaluate(model, data_loader_test, device=device)

# Plot the results
plt.plot(hybrid_results)
plt.xlabel('Epoch')
plt.ylabel('Hybrid Technique (1 for Adam, 0 for SGD)')
plt.title('Hybrid Optimization Technique')
plt.show()

Epoch: [0]  [  0/104]  eta: 0:03:56  lr: 0.000011  loss: 3.3583 (3.3583)  loss_classifier: 2.5736 (2.5736)  loss_box_reg: 0.4333 (0.4333)  loss_objectness: 0.3133 (0.3133)  loss_rpn_box_reg: 0.0381 (0.0381)  time: 2.2760  data: 0.6913  max mem: 6189
Epoch: [0]  [ 10/104]  eta: 0:01:57  lr: 0.000108  loss: 2.3279 (2.1386)  loss_classifier: 1.5769 (1.5182)  loss_box_reg: 0.3590 (0.3778)  loss_objectness: 0.1724 (0.2194)  loss_rpn_box_reg: 0.0200 (0.0231)  time: 1.2486  data: 0.0965  max mem: 6507
Epoch: [0]  [ 20/104]  eta: 0:01:39  lr: 0.000205  loss: 1.2090 (1.6584)  loss_classifier: 0.6640 (1.1010)  loss_box_reg: 0.3590 (0.3867)  loss_objectness: 0.0846 (0.1480)  loss_rpn_box_reg: 0.0191 (0.0227)  time: 1.1307  data: 0.0314  max mem: 6507
Epoch: [0]  [ 30/104]  eta: 0:01:26  lr: 0.000302  loss: 0.9238 (1.3821)  loss_classifier: 0.4708 (0.8761)  loss_box_reg: 0.3461 (0.3638)  loss_objectness: 0.0666 (0.1216)  loss_rpn_box_reg: 0.0139 (0.0207)  time: 1.1243  data: 0.0262  max mem: 6507
Epoch: [0]  [ 40/104]  eta: 0:01:13  lr: 0.000399  loss: 0.8790 (1.2982)  loss_classifier: 0.4517 (0.7931)  loss_box_reg: 0.3577 (0.3717)  loss_objectness: 0.0476 (0.1117)  loss_rpn_box_reg: 0.0140 (0.0217)  time: 1.1121  data: 0.0269  max mem: 6507
Epoch: [0]  [ 50/104]  eta: 0:01:01  lr: 0.000496  loss: 0.9410 (1.2036)  loss_classifier: 0.4517 (0.7171)  loss_box_reg: 0.3279 (0.3578)  loss_objectness: 0.0535 (0.1052)  loss_rpn_box_reg: 0.0218 (0.0235)  time: 1.0731  data: 0.0233  max mem: 6507
Epoch: [0]  [ 60/104]  eta: 0:00:49  lr: 0.000593  loss: 0.8082 (1.1516)  loss_classifier: 0.3806 (0.6660)  loss_box_reg: 0.3105 (0.3600)  loss_objectness: 0.0697 (0.1014)  loss_rpn_box_reg: 0.0244 (0.0241)  time: 1.0451  data: 0.0211  max mem: 6507
Epoch: [0]  [ 70/104]  eta: 0:00:37  lr: 0.000690  loss: 0.6632 (1.0734)  loss_classifier: 0.3152 (0.6125)  loss_box_reg: 0.2753 (0.3432)  loss_objectness: 0.0606 (0.0945)  loss_rpn_box_reg: 0.0175 (0.0231)  time: 1.0392  data: 0.0225  max mem: 6507
Epoch: [0]  [ 80/104]  eta: 0:00:26  lr: 0.000787  loss: 0.5934 (1.0429)  loss_classifier: 0.3131 (0.5863)  loss_box_reg: 0.2408 (0.3441)  loss_objectness: 0.0489 (0.0897)  loss_rpn_box_reg: 0.0122 (0.0228)  time: 1.0399  data: 0.0216  max mem: 6507
Epoch: [0]  [ 90/104]  eta: 0:00:15  lr: 0.000884  loss: 0.8766 (1.0211)  loss_classifier: 0.3554 (0.5630)  loss_box_reg: 0.3787 (0.3491)  loss_objectness: 0.0441 (0.0859)  loss_rpn_box_reg: 0.0196 (0.0231)  time: 1.0379  data: 0.0204  max mem: 6507
Epoch: [0]  [100/104]  eta: 0:00:04  lr: 0.000981  loss: 0.7816 (0.9918)  loss_classifier: 0.3472 (0.5380)  loss_box_reg: 0.3829 (0.3490)  loss_objectness: 0.0430 (0.0822)  loss_rpn_box_reg: 0.0200 (0.0225)  time: 1.0439  data: 0.0200  max mem: 6507
Epoch: [0]  [103/104]  eta: 0:00:01  lr: 0.001000  loss: 0.6924 (0.9838)  loss_classifier: 0.3055 (0.5306)  loss_box_reg: 0.3446 (0.3494)  loss_objectness: 0.0425 (0.0814)  loss_rpn_box_reg: 0.0163 (0.0224)  time: 1.0474  data: 0.0201  max mem: 6507
Epoch: [0] Total time: 0:01:53 (1.0872 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:34  model_time: 0.5353 (0.5353)  evaluator_time: 0.0408 (0.0408)  time: 1.3204  data: 0.7312  max mem: 6507
Test:  [25/26]  eta: 0:00:00  model_time: 0.4609 (0.4643)  evaluator_time: 0.0232 (0.0326)  time: 0.5161  data: 0.0198  max mem: 6507
Test: Total time: 0:00:14 (0.5556 s / it)
Averaged stats: model_time: 0.4609 (0.4643)  evaluator_time: 0.0232 (0.0326)
Accumulating evaluation results...
DONE (t=0.25s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.098
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.314
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.024
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.066
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.080
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.122
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.073
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.178
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.223
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.246
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.178
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.221
Epoch: [1]  [  0/104]  eta: 0:03:22  lr: 0.001000  loss: 0.5595 (0.5595)  loss_classifier: 0.2305 (0.2305)  loss_box_reg: 0.2976 (0.2976)  loss_objectness: 0.0189 (0.0189)  loss_rpn_box_reg: 0.0124 (0.0124)  time: 1.9507  data: 0.7740  max mem: 6507
Epoch: [1]  [ 10/104]  eta: 0:01:51  lr: 0.001000  loss: 0.9238 (0.8926)  loss_classifier: 0.3809 (0.3904)  loss_box_reg: 0.4200 (0.4017)  loss_objectness: 0.0522 (0.0652)  loss_rpn_box_reg: 0.0273 (0.0353)  time: 1.1860  data: 0.0873  max mem: 6508
Epoch: [1]  [ 20/104]  eta: 0:01:34  lr: 0.001000  loss: 0.9238 (0.9645)  loss_classifier: 0.4395 (0.4409)  loss_box_reg: 0.3396 (0.3422)  loss_objectness: 0.0792 (0.1439)  loss_rpn_box_reg: 0.0273 (0.0375)  time: 1.0846  data: 0.0192  max mem: 6508
Epoch: [1]  [ 30/104]  eta: 0:01:21  lr: 0.001000  loss: 0.8937 (0.9598)  loss_classifier: 0.4452 (0.4378)  loss_box_reg: 0.2570 (0.3153)  loss_objectness: 0.1600 (0.1692)  loss_rpn_box_reg: 0.0256 (0.0375)  time: 1.0453  data: 0.0202  max mem: 6508
Epoch: [1]  [ 40/104]  eta: 0:01:08  lr: 0.001000  loss: 0.8828 (0.9770)  loss_classifier: 0.4144 (0.4524)  loss_box_reg: 0.2087 (0.2943)  loss_objectness: 0.1780 (0.1906)  loss_rpn_box_reg: 0.0408 (0.0397)  time: 1.0205  data: 0.0201  max mem: 6508
Epoch: [1]  [ 50/104]  eta: 0:00:57  lr: 0.001000  loss: 0.8035 (0.9467)  loss_classifier: 0.3222 (0.4411)  loss_box_reg: 0.1765 (0.2818)  loss_objectness: 0.1685 (0.1855)  loss_rpn_box_reg: 0.0305 (0.0384)  time: 1.0066  data: 0.0203  max mem: 6508
Epoch: [1]  [ 60/104]  eta: 0:00:46  lr: 0.001000  loss: 0.7846 (0.9380)  loss_classifier: 0.3764 (0.4374)  loss_box_reg: 0.2107 (0.2764)  loss_objectness: 0.1683 (0.1855)  loss_rpn_box_reg: 0.0292 (0.0387)  time: 1.0037  data: 0.0221  max mem: 6508
Epoch: [1]  [ 70/104]  eta: 0:00:35  lr: 0.001000  loss: 0.7665 (0.9111)  loss_classifier: 0.3365 (0.4234)  loss_box_reg: 0.1780 (0.2724)  loss_objectness: 0.1311 (0.1771)  loss_rpn_box_reg: 0.0338 (0.0382)  time: 1.0144  data: 0.0230  max mem: 6508
Epoch: [1]  [ 80/104]  eta: 0:00:25  lr: 0.001000  loss: 0.7328 (0.9069)  loss_classifier: 0.3365 (0.4278)  loss_box_reg: 0.2070 (0.2725)  loss_objectness: 0.1122 (0.1698)  loss_rpn_box_reg: 0.0210 (0.0367)  time: 1.0243  data: 0.0219  max mem: 6508
Epoch: [1]  [ 90/104]  eta: 0:00:14  lr: 0.001000  loss: 0.8004 (0.9005)  loss_classifier: 0.3582 (0.4239)  loss_box_reg: 0.2739 (0.2731)  loss_objectness: 0.1272 (0.1678)  loss_rpn_box_reg: 0.0222 (0.0357)  time: 1.0265  data: 0.0206  max mem: 6508
Epoch: [1]  [100/104]  eta: 0:00:04  lr: 0.001000  loss: 0.8303 (0.8894)  loss_classifier: 0.3423 (0.4180)  loss_box_reg: 0.2363 (0.2700)  loss_objectness: 0.1451 (0.1664)  loss_rpn_box_reg: 0.0222 (0.0350)  time: 1.0341  data: 0.0204  max mem: 6508
Epoch: [1]  [103/104]  eta: 0:00:01  lr: 0.001000  loss: 0.8303 (0.8839)  loss_classifier: 0.3094 (0.4124)  loss_box_reg: 0.2845 (0.2710)  loss_objectness: 0.1452 (0.1656)  loss_rpn_box_reg: 0.0225 (0.0349)  time: 1.0314  data: 0.0192  max mem: 6508
Epoch: [1] Total time: 0:01:48 (1.0433 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:30  model_time: 0.4873 (0.4873)  evaluator_time: 0.0132 (0.0132)  time: 1.1813  data: 0.6721  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4349 (0.4359)  evaluator_time: 0.0175 (0.0435)  time: 0.5049  data: 0.0208  max mem: 6508
Test: Total time: 0:00:13 (0.5350 s / it)
Averaged stats: model_time: 0.4349 (0.4359)  evaluator_time: 0.0175 (0.0435)
Accumulating evaluation results...
DONE (t=0.14s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.032
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.112
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.004
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.028
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.063
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.007
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.056
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.101
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.115
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.154
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
Epoch: [2]  [  0/104]  eta: 0:03:12  lr: 0.001000  loss: 0.5037 (0.5037)  loss_classifier: 0.2505 (0.2505)  loss_box_reg: 0.1813 (0.1813)  loss_objectness: 0.0628 (0.0628)  loss_rpn_box_reg: 0.0091 (0.0091)  time: 1.8531  data: 0.7893  max mem: 6508
Epoch: [2]  [ 10/104]  eta: 0:01:45  lr: 0.001000  loss: 0.7679 (0.7480)  loss_classifier: 0.2630 (0.3047)  loss_box_reg: 0.3321 (0.2946)  loss_objectness: 0.1093 (0.1169)  loss_rpn_box_reg: 0.0268 (0.0318)  time: 1.1201  data: 0.0887  max mem: 6508
Epoch: [2]  [ 20/104]  eta: 0:01:31  lr: 0.001000  loss: 0.7679 (0.8105)  loss_classifier: 0.3391 (0.3457)  loss_box_reg: 0.3321 (0.3236)  loss_objectness: 0.1086 (0.1076)  loss_rpn_box_reg: 0.0280 (0.0336)  time: 1.0487  data: 0.0201  max mem: 6508
Epoch: [2]  [ 30/104]  eta: 0:01:19  lr: 0.001000  loss: 0.6997 (0.7948)  loss_classifier: 0.3233 (0.3411)  loss_box_reg: 0.2844 (0.3074)  loss_objectness: 0.1138 (0.1135)  loss_rpn_box_reg: 0.0254 (0.0328)  time: 1.0460  data: 0.0213  max mem: 6508
Epoch: [2]  [ 40/104]  eta: 0:01:07  lr: 0.001000  loss: 0.6524 (0.7697)  loss_classifier: 0.3024 (0.3320)  loss_box_reg: 0.2380 (0.3007)  loss_objectness: 0.0968 (0.1053)  loss_rpn_box_reg: 0.0232 (0.0317)  time: 1.0326  data: 0.0201  max mem: 6508
Epoch: [2]  [ 50/104]  eta: 0:00:57  lr: 0.001000  loss: 0.6844 (0.7759)  loss_classifier: 0.3101 (0.3390)  loss_box_reg: 0.2446 (0.3037)  loss_objectness: 0.0753 (0.1016)  loss_rpn_box_reg: 0.0267 (0.0317)  time: 1.0467  data: 0.0246  max mem: 6508
Epoch: [2]  [ 60/104]  eta: 0:00:46  lr: 0.001000  loss: 0.6844 (0.7725)  loss_classifier: 0.3198 (0.3358)  loss_box_reg: 0.2572 (0.3075)  loss_objectness: 0.0772 (0.0983)  loss_rpn_box_reg: 0.0240 (0.0309)  time: 1.0442  data: 0.0251  max mem: 6508
Epoch: [2]  [ 70/104]  eta: 0:00:35  lr: 0.001000  loss: 0.7528 (0.7731)  loss_classifier: 0.3344 (0.3377)  loss_box_reg: 0.3064 (0.3110)  loss_objectness: 0.0596 (0.0942)  loss_rpn_box_reg: 0.0235 (0.0301)  time: 1.0312  data: 0.0238  max mem: 6508
Epoch: [2]  [ 80/104]  eta: 0:00:25  lr: 0.001000  loss: 0.8539 (0.7948)  loss_classifier: 0.3837 (0.3514)  loss_box_reg: 0.3581 (0.3217)  loss_objectness: 0.0577 (0.0909)  loss_rpn_box_reg: 0.0265 (0.0308)  time: 1.0315  data: 0.0231  max mem: 6508
Epoch: [2]  [ 90/104]  eta: 0:00:14  lr: 0.001000  loss: 0.8539 (0.7890)  loss_classifier: 0.3838 (0.3490)  loss_box_reg: 0.3581 (0.3225)  loss_objectness: 0.0537 (0.0868)  loss_rpn_box_reg: 0.0274 (0.0306)  time: 1.0237  data: 0.0194  max mem: 6508
Epoch: [2]  [100/104]  eta: 0:00:04  lr: 0.001000  loss: 0.6873 (0.7845)  loss_classifier: 0.2968 (0.3464)  loss_box_reg: 0.2919 (0.3230)  loss_objectness: 0.0533 (0.0849)  loss_rpn_box_reg: 0.0234 (0.0302)  time: 1.0292  data: 0.0194  max mem: 6508
Epoch: [2]  [103/104]  eta: 0:00:01  lr: 0.001000  loss: 0.6611 (0.7760)  loss_classifier: 0.2968 (0.3432)  loss_box_reg: 0.2749 (0.3188)  loss_objectness: 0.0537 (0.0843)  loss_rpn_box_reg: 0.0180 (0.0297)  time: 1.0298  data: 0.0200  max mem: 6508
Epoch: [2] Total time: 0:01:48 (1.0476 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:32  model_time: 0.5254 (0.5254)  evaluator_time: 0.0422 (0.0422)  time: 1.2660  data: 0.6837  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4378 (0.4407)  evaluator_time: 0.0279 (0.0307)  time: 0.4987  data: 0.0235  max mem: 6508
Test: Total time: 0:00:13 (0.5316 s / it)
Averaged stats: model_time: 0.4378 (0.4407)  evaluator_time: 0.0279 (0.0307)
Accumulating evaluation results...
DONE (t=0.17s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.095
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.293
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.024
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.101
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.115
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.006
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.045
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.174
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.226
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.235
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.278
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.090
Epoch: [3]  [  0/104]  eta: 0:03:13  lr: 0.001000  loss: 0.9499 (0.9499)  loss_classifier: 0.3417 (0.3417)  loss_box_reg: 0.4621 (0.4621)  loss_objectness: 0.1011 (0.1011)  loss_rpn_box_reg: 0.0450 (0.0450)  time: 1.8584  data: 0.8103  max mem: 6508
Epoch: [3]  [ 10/104]  eta: 0:01:47  lr: 0.001000  loss: 0.8302 (0.8235)  loss_classifier: 0.3417 (0.3547)  loss_box_reg: 0.3616 (0.3747)  loss_objectness: 0.0448 (0.0643)  loss_rpn_box_reg: 0.0273 (0.0298)  time: 1.1402  data: 0.0937  max mem: 6508
Epoch: [3]  [ 20/104]  eta: 0:01:32  lr: 0.001000  loss: 0.6848 (0.7629)  loss_classifier: 0.3111 (0.3349)  loss_box_reg: 0.3178 (0.3431)  loss_objectness: 0.0435 (0.0587)  loss_rpn_box_reg: 0.0234 (0.0263)  time: 1.0628  data: 0.0214  max mem: 6508
Epoch: [3]  [ 30/104]  eta: 0:01:19  lr: 0.001000  loss: 0.7425 (0.7528)  loss_classifier: 0.3458 (0.3322)  loss_box_reg: 0.3276 (0.3383)  loss_objectness: 0.0473 (0.0571)  loss_rpn_box_reg: 0.0217 (0.0252)  time: 1.0461  data: 0.0202  max mem: 6508
Epoch: [3]  [ 40/104]  eta: 0:01:08  lr: 0.001000  loss: 0.7538 (0.7365)  loss_classifier: 0.2820 (0.3206)  loss_box_reg: 0.3276 (0.3360)  loss_objectness: 0.0410 (0.0545)  loss_rpn_box_reg: 0.0217 (0.0255)  time: 1.0305  data: 0.0193  max mem: 6508
Epoch: [3]  [ 50/104]  eta: 0:00:57  lr: 0.001000  loss: 0.6383 (0.7350)  loss_classifier: 0.2502 (0.3181)  loss_box_reg: 0.3137 (0.3374)  loss_objectness: 0.0410 (0.0530)  loss_rpn_box_reg: 0.0182 (0.0266)  time: 1.0231  data: 0.0199  max mem: 6508
Epoch: [3]  [ 60/104]  eta: 0:00:46  lr: 0.001000  loss: 0.6383 (0.7166)  loss_classifier: 0.2553 (0.3114)  loss_box_reg: 0.2984 (0.3277)  loss_objectness: 0.0440 (0.0519)  loss_rpn_box_reg: 0.0172 (0.0255)  time: 1.0162  data: 0.0202  max mem: 6508
Epoch: [3]  [ 70/104]  eta: 0:00:35  lr: 0.001000  loss: 0.5951 (0.7024)  loss_classifier: 0.2605 (0.3047)  loss_box_reg: 0.2664 (0.3204)  loss_objectness: 0.0458 (0.0515)  loss_rpn_box_reg: 0.0172 (0.0258)  time: 1.0147  data: 0.0198  max mem: 6508
Epoch: [3]  [ 80/104]  eta: 0:00:25  lr: 0.001000  loss: 0.6758 (0.7072)  loss_classifier: 0.2742 (0.3053)  loss_box_reg: 0.3144 (0.3232)  loss_objectness: 0.0472 (0.0520)  loss_rpn_box_reg: 0.0216 (0.0266)  time: 1.0216  data: 0.0210  max mem: 6508
Epoch: [3]  [ 90/104]  eta: 0:00:14  lr: 0.001000  loss: 0.6085 (0.6951)  loss_classifier: 0.2477 (0.2973)  loss_box_reg: 0.3144 (0.3206)  loss_objectness: 0.0452 (0.0510)  loss_rpn_box_reg: 0.0232 (0.0263)  time: 1.0288  data: 0.0217  max mem: 6508
Epoch: [3]  [100/104]  eta: 0:00:04  lr: 0.001000  loss: 0.5904 (0.6938)  loss_classifier: 0.2427 (0.2959)  loss_box_reg: 0.2899 (0.3212)  loss_objectness: 0.0452 (0.0511)  loss_rpn_box_reg: 0.0188 (0.0257)  time: 1.0302  data: 0.0201  max mem: 6508
Epoch: [3]  [103/104]  eta: 0:00:01  lr: 0.001000  loss: 0.6085 (0.6953)  loss_classifier: 0.2726 (0.2968)  loss_box_reg: 0.3100 (0.3219)  loss_objectness: 0.0410 (0.0507)  loss_rpn_box_reg: 0.0193 (0.0259)  time: 1.0272  data: 0.0186  max mem: 6508
Epoch: [3] Total time: 0:01:48 (1.0411 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:32  model_time: 0.5188 (0.5188)  evaluator_time: 0.0565 (0.0565)  time: 1.2508  data: 0.6631  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4385 (0.4402)  evaluator_time: 0.0283 (0.0381)  time: 0.5035  data: 0.0237  max mem: 6508
Test: Total time: 0:00:13 (0.5367 s / it)
Averaged stats: model_time: 0.4385 (0.4402)  evaluator_time: 0.0283 (0.0381)
Accumulating evaluation results...
DONE (t=0.17s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.173
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.455
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.059
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.190
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.208
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.008
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.071
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.259
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.323
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.334
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.374
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.060
Epoch: [4]  [  0/104]  eta: 0:03:07  lr: 0.001000  loss: 0.9558 (0.9558)  loss_classifier: 0.4930 (0.4930)  loss_box_reg: 0.3486 (0.3486)  loss_objectness: 0.0946 (0.0946)  loss_rpn_box_reg: 0.0195 (0.0195)  time: 1.8037  data: 0.7351  max mem: 6508
Epoch: [4]  [ 10/104]  eta: 0:01:45  lr: 0.001000  loss: 0.7101 (0.6399)  loss_classifier: 0.2657 (0.2640)  loss_box_reg: 0.3435 (0.3079)  loss_objectness: 0.0472 (0.0477)  loss_rpn_box_reg: 0.0195 (0.0202)  time: 1.1204  data: 0.0832  max mem: 6508
Epoch: [4]  [ 20/104]  eta: 0:01:32  lr: 0.001000  loss: 0.6309 (0.6270)  loss_classifier: 0.2392 (0.2552)  loss_box_reg: 0.2930 (0.3038)  loss_objectness: 0.0401 (0.0461)  loss_rpn_box_reg: 0.0210 (0.0220)  time: 1.0610  data: 0.0199  max mem: 6508
Epoch: [4]  [ 30/104]  eta: 0:01:19  lr: 0.001000  loss: 0.6323 (0.6352)  loss_classifier: 0.2409 (0.2580)  loss_box_reg: 0.2970 (0.3073)  loss_objectness: 0.0392 (0.0461)  loss_rpn_box_reg: 0.0210 (0.0238)  time: 1.0517  data: 0.0202  max mem: 6508
Epoch: [4]  [ 40/104]  eta: 0:01:08  lr: 0.001000  loss: 0.6639 (0.6324)  loss_classifier: 0.2750 (0.2566)  loss_box_reg: 0.2970 (0.3055)  loss_objectness: 0.0370 (0.0458)  loss_rpn_box_reg: 0.0238 (0.0244)  time: 1.0370  data: 0.0202  max mem: 6508
Epoch: [4]  [ 50/104]  eta: 0:00:57  lr: 0.001000  loss: 0.5750 (0.6164)  loss_classifier: 0.2283 (0.2477)  loss_box_reg: 0.2813 (0.3012)  loss_objectness: 0.0317 (0.0434)  loss_rpn_box_reg: 0.0209 (0.0241)  time: 1.0452  data: 0.0273  max mem: 6508
Epoch: [4]  [ 60/104]  eta: 0:00:46  lr: 0.001000  loss: 0.6330 (0.6248)  loss_classifier: 0.2227 (0.2506)  loss_box_reg: 0.2909 (0.3078)  loss_objectness: 0.0316 (0.0428)  loss_rpn_box_reg: 0.0194 (0.0236)  time: 1.0604  data: 0.0331  max mem: 6508
Epoch: [4]  [ 70/104]  eta: 0:00:35  lr: 0.001000  loss: 0.6562 (0.6291)  loss_classifier: 0.2589 (0.2515)  loss_box_reg: 0.3541 (0.3110)  loss_objectness: 0.0412 (0.0432)  loss_rpn_box_reg: 0.0197 (0.0235)  time: 1.0427  data: 0.0264  max mem: 6508
Epoch: [4]  [ 80/104]  eta: 0:00:25  lr: 0.001000  loss: 0.6114 (0.6299)  loss_classifier: 0.2353 (0.2526)  loss_box_reg: 0.2933 (0.3100)  loss_objectness: 0.0382 (0.0435)  loss_rpn_box_reg: 0.0213 (0.0237)  time: 1.0201  data: 0.0193  max mem: 6508
Epoch: [4]  [ 90/104]  eta: 0:00:14  lr: 0.001000  loss: 0.5727 (0.6264)  loss_classifier: 0.2379 (0.2512)  loss_box_reg: 0.2812 (0.3085)  loss_objectness: 0.0370 (0.0428)  loss_rpn_box_reg: 0.0217 (0.0239)  time: 1.0289  data: 0.0199  max mem: 6508
Epoch: [4]  [100/104]  eta: 0:00:04  lr: 0.001000  loss: 0.5472 (0.6211)  loss_classifier: 0.2376 (0.2497)  loss_box_reg: 0.2812 (0.3055)  loss_objectness: 0.0378 (0.0427)  loss_rpn_box_reg: 0.0183 (0.0233)  time: 1.0302  data: 0.0198  max mem: 6508
Epoch: [4]  [103/104]  eta: 0:00:01  lr: 0.001000  loss: 0.5416 (0.6168)  loss_classifier: 0.2112 (0.2477)  loss_box_reg: 0.2812 (0.3037)  loss_objectness: 0.0353 (0.0423)  loss_rpn_box_reg: 0.0182 (0.0230)  time: 1.0283  data: 0.0193  max mem: 6508
Epoch: [4] Total time: 0:01:49 (1.0498 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:32  model_time: 0.5054 (0.5054)  evaluator_time: 0.0618 (0.0618)  time: 1.2496  data: 0.6658  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4377 (0.4409)  evaluator_time: 0.0372 (0.0436)  time: 0.5433  data: 0.0468  max mem: 6508
Test: Total time: 0:00:14 (0.5649 s / it)
Averaged stats: model_time: 0.4377 (0.4409)  evaluator_time: 0.0372 (0.0436)
Accumulating evaluation results...
DONE (t=0.47s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.225
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.524
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.150
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.170
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.237
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.108
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.105
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.315
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.371
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.350
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.373
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.159
Epoch: [5]  [  0/104]  eta: 0:03:42  lr: 0.001000  loss: 0.4680 (0.4680)  loss_classifier: 0.1478 (0.1478)  loss_box_reg: 0.2861 (0.2861)  loss_objectness: 0.0217 (0.0217)  loss_rpn_box_reg: 0.0124 (0.0124)  time: 2.1373  data: 0.9660  max mem: 6508
Epoch: [5]  [ 10/104]  eta: 0:01:50  lr: 0.001000  loss: 0.4746 (0.5283)  loss_classifier: 0.2071 (0.2146)  loss_box_reg: 0.2416 (0.2675)  loss_objectness: 0.0250 (0.0293)  loss_rpn_box_reg: 0.0146 (0.0170)  time: 1.1784  data: 0.1125  max mem: 6508
Epoch: [5]  [ 20/104]  eta: 0:01:35  lr: 0.001000  loss: 0.5318 (0.5453)  loss_classifier: 0.2175 (0.2221)  loss_box_reg: 0.2739 (0.2739)  loss_objectness: 0.0252 (0.0321)  loss_rpn_box_reg: 0.0154 (0.0172)  time: 1.0828  data: 0.0252  max mem: 6508
Epoch: [5]  [ 30/104]  eta: 0:01:21  lr: 0.001000  loss: 0.6038 (0.5736)  loss_classifier: 0.2350 (0.2375)  loss_box_reg: 0.2890 (0.2788)  loss_objectness: 0.0325 (0.0391)  loss_rpn_box_reg: 0.0162 (0.0182)  time: 1.0645  data: 0.0229  max mem: 6508
Epoch: [5]  [ 40/104]  eta: 0:01:09  lr: 0.001000  loss: 0.5810 (0.5703)  loss_classifier: 0.2350 (0.2362)  loss_box_reg: 0.2642 (0.2782)  loss_objectness: 0.0323 (0.0385)  loss_rpn_box_reg: 0.0140 (0.0174)  time: 1.0346  data: 0.0205  max mem: 6508
Epoch: [5]  [ 50/104]  eta: 0:00:57  lr: 0.001000  loss: 0.5266 (0.5752)  loss_classifier: 0.2094 (0.2369)  loss_box_reg: 0.2642 (0.2821)  loss_objectness: 0.0310 (0.0368)  loss_rpn_box_reg: 0.0175 (0.0193)  time: 1.0244  data: 0.0204  max mem: 6508
Epoch: [5]  [ 60/104]  eta: 0:00:46  lr: 0.001000  loss: 0.5092 (0.5591)  loss_classifier: 0.1943 (0.2278)  loss_box_reg: 0.2546 (0.2764)  loss_objectness: 0.0306 (0.0361)  loss_rpn_box_reg: 0.0192 (0.0188)  time: 1.0229  data: 0.0221  max mem: 6508
Epoch: [5]  [ 70/104]  eta: 0:00:36  lr: 0.001000  loss: 0.4859 (0.5504)  loss_classifier: 0.1813 (0.2209)  loss_box_reg: 0.2525 (0.2763)  loss_objectness: 0.0248 (0.0347)  loss_rpn_box_reg: 0.0164 (0.0185)  time: 1.0264  data: 0.0248  max mem: 6508
Epoch: [5]  [ 80/104]  eta: 0:00:25  lr: 0.001000  loss: 0.5286 (0.5521)  loss_classifier: 0.1831 (0.2182)  loss_box_reg: 0.2902 (0.2812)  loss_objectness: 0.0236 (0.0337)  loss_rpn_box_reg: 0.0197 (0.0190)  time: 1.0305  data: 0.0253  max mem: 6508
Epoch: [5]  [ 90/104]  eta: 0:00:14  lr: 0.001000  loss: 0.5525 (0.5546)  loss_classifier: 0.1898 (0.2181)  loss_box_reg: 0.3005 (0.2828)  loss_objectness: 0.0242 (0.0339)  loss_rpn_box_reg: 0.0217 (0.0198)  time: 1.0311  data: 0.0222  max mem: 6508
Epoch: [5]  [100/104]  eta: 0:00:04  lr: 0.001000  loss: 0.5126 (0.5501)  loss_classifier: 0.1921 (0.2168)  loss_box_reg: 0.2651 (0.2804)  loss_objectness: 0.0218 (0.0332)  loss_rpn_box_reg: 0.0170 (0.0197)  time: 1.0316  data: 0.0204  max mem: 6508
Epoch: [5]  [103/104]  eta: 0:00:01  lr: 0.001000  loss: 0.4816 (0.5490)  loss_classifier: 0.1829 (0.2162)  loss_box_reg: 0.2651 (0.2799)  loss_objectness: 0.0218 (0.0333)  loss_rpn_box_reg: 0.0170 (0.0196)  time: 1.0310  data: 0.0202  max mem: 6508
Epoch: [5] Total time: 0:01:49 (1.0521 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:35  model_time: 0.4859 (0.4859)  evaluator_time: 0.0420 (0.0420)  time: 1.3520  data: 0.8006  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4382 (0.4394)  evaluator_time: 0.0253 (0.0298)  time: 0.5024  data: 0.0250  max mem: 6508
Test: Total time: 0:00:13 (0.5345 s / it)
Averaged stats: model_time: 0.4382 (0.4394)  evaluator_time: 0.0253 (0.0298)
Accumulating evaluation results...
DONE (t=0.17s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.290
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.654
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.201
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.218
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.335
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.190
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.123
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.375
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.441
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.376
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.504
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.237
Epoch: [6]  [  0/104]  eta: 0:03:03  lr: 0.001000  loss: 0.2004 (0.2004)  loss_classifier: 0.0850 (0.0850)  loss_box_reg: 0.0903 (0.0903)  loss_objectness: 0.0186 (0.0186)  loss_rpn_box_reg: 0.0066 (0.0066)  time: 1.7673  data: 0.6507  max mem: 6508
Epoch: [6]  [ 10/104]  eta: 0:01:46  lr: 0.001000  loss: 0.3790 (0.4026)  loss_classifier: 0.1573 (0.1662)  loss_box_reg: 0.1895 (0.2015)  loss_objectness: 0.0203 (0.0207)  loss_rpn_box_reg: 0.0126 (0.0142)  time: 1.1316  data: 0.0830  max mem: 6508
Epoch: [6]  [ 20/104]  eta: 0:01:31  lr: 0.001000  loss: 0.3790 (0.4043)  loss_classifier: 0.1573 (0.1620)  loss_box_reg: 0.2113 (0.2048)  loss_objectness: 0.0220 (0.0234)  loss_rpn_box_reg: 0.0126 (0.0142)  time: 1.0554  data: 0.0230  max mem: 6508
Epoch: [6]  [ 30/104]  eta: 0:01:19  lr: 0.001000  loss: 0.3963 (0.4200)  loss_classifier: 0.1639 (0.1660)  loss_box_reg: 0.2222 (0.2154)  loss_objectness: 0.0230 (0.0232)  loss_rpn_box_reg: 0.0141 (0.0154)  time: 1.0335  data: 0.0191  max mem: 6508
Epoch: [6]  [ 40/104]  eta: 0:01:07  lr: 0.001000  loss: 0.4557 (0.4215)  loss_classifier: 0.1723 (0.1667)  loss_box_reg: 0.2258 (0.2188)  loss_objectness: 0.0183 (0.0212)  loss_rpn_box_reg: 0.0150 (0.0149)  time: 1.0207  data: 0.0190  max mem: 6508
Epoch: [6]  [ 50/104]  eta: 0:00:56  lr: 0.001000  loss: 0.4646 (0.4297)  loss_classifier: 0.1774 (0.1703)  loss_box_reg: 0.2445 (0.2235)  loss_objectness: 0.0139 (0.0200)  loss_rpn_box_reg: 0.0146 (0.0160)  time: 1.0178  data: 0.0224  max mem: 6508
Epoch: [6]  [ 60/104]  eta: 0:00:45  lr: 0.001000  loss: 0.4693 (0.4323)  loss_classifier: 0.1746 (0.1713)  loss_box_reg: 0.2554 (0.2241)  loss_objectness: 0.0151 (0.0204)  loss_rpn_box_reg: 0.0173 (0.0166)  time: 1.0160  data: 0.0231  max mem: 6508
Epoch: [6]  [ 70/104]  eta: 0:00:35  lr: 0.001000  loss: 0.4249 (0.4347)  loss_classifier: 0.1655 (0.1703)  loss_box_reg: 0.2321 (0.2279)  loss_objectness: 0.0179 (0.0199)  loss_rpn_box_reg: 0.0166 (0.0166)  time: 1.0130  data: 0.0207  max mem: 6508
Epoch: [6]  [ 80/104]  eta: 0:00:24  lr: 0.001000  loss: 0.4432 (0.4389)  loss_classifier: 0.1613 (0.1696)  loss_box_reg: 0.2495 (0.2335)  loss_objectness: 0.0150 (0.0193)  loss_rpn_box_reg: 0.0130 (0.0164)  time: 1.0174  data: 0.0218  max mem: 6508
Epoch: [6]  [ 90/104]  eta: 0:00:14  lr: 0.001000  loss: 0.4432 (0.4411)  loss_classifier: 0.1654 (0.1694)  loss_box_reg: 0.2633 (0.2356)  loss_objectness: 0.0155 (0.0195)  loss_rpn_box_reg: 0.0130 (0.0166)  time: 1.0281  data: 0.0234  max mem: 6508
Epoch: [6]  [100/104]  eta: 0:00:04  lr: 0.001000  loss: 0.3908 (0.4347)  loss_classifier: 0.1517 (0.1675)  loss_box_reg: 0.2001 (0.2324)  loss_objectness: 0.0155 (0.0190)  loss_rpn_box_reg: 0.0096 (0.0158)  time: 1.0319  data: 0.0225  max mem: 6508
Epoch: [6]  [103/104]  eta: 0:00:01  lr: 0.001000  loss: 0.3908 (0.4337)  loss_classifier: 0.1499 (0.1669)  loss_box_reg: 0.2096 (0.2322)  loss_objectness: 0.0147 (0.0188)  loss_rpn_box_reg: 0.0106 (0.0158)  time: 1.0323  data: 0.0220  max mem: 6508
Epoch: [6] Total time: 0:01:47 (1.0369 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:33  model_time: 0.5286 (0.5286)  evaluator_time: 0.0611 (0.0611)  time: 1.2888  data: 0.6926  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4329 (0.4359)  evaluator_time: 0.0202 (0.0268)  time: 0.4862  data: 0.0207  max mem: 6508
Test: Total time: 0:00:13 (0.5192 s / it)
Averaged stats: model_time: 0.4329 (0.4359)  evaluator_time: 0.0202 (0.0268)
Accumulating evaluation results...
DONE (t=0.16s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.347
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.691
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.295
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.333
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.406
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.231
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.153
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.433
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.501
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.413
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.553
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.329
Epoch: [7]  [  0/104]  eta: 0:03:26  lr: 0.001000  loss: 0.5793 (0.5793)  loss_classifier: 0.2718 (0.2718)  loss_box_reg: 0.2628 (0.2628)  loss_objectness: 0.0261 (0.0261)  loss_rpn_box_reg: 0.0186 (0.0186)  time: 1.9895  data: 0.9524  max mem: 6508
Epoch: [7]  [ 10/104]  eta: 0:01:46  lr: 0.001000  loss: 0.3872 (0.4365)  loss_classifier: 0.1535 (0.1738)  loss_box_reg: 0.2187 (0.2288)  loss_objectness: 0.0147 (0.0176)  loss_rpn_box_reg: 0.0158 (0.0163)  time: 1.1289  data: 0.1037  max mem: 6508
Epoch: [7]  [ 20/104]  eta: 0:01:32  lr: 0.001000  loss: 0.3806 (0.4099)  loss_classifier: 0.1420 (0.1582)  loss_box_reg: 0.2103 (0.2188)  loss_objectness: 0.0147 (0.0185)  loss_rpn_box_reg: 0.0118 (0.0143)  time: 1.0520  data: 0.0194  max mem: 6508
Epoch: [7]  [ 30/104]  eta: 0:01:19  lr: 0.001000  loss: 0.4133 (0.4081)  loss_classifier: 0.1405 (0.1565)  loss_box_reg: 0.2212 (0.2168)  loss_objectness: 0.0172 (0.0196)  loss_rpn_box_reg: 0.0121 (0.0152)  time: 1.0527  data: 0.0213  max mem: 6508
Epoch: [7]  [ 40/104]  eta: 0:01:08  lr: 0.001000  loss: 0.3974 (0.3959)  loss_classifier: 0.1418 (0.1503)  loss_box_reg: 0.1994 (0.2121)  loss_objectness: 0.0180 (0.0193)  loss_rpn_box_reg: 0.0115 (0.0142)  time: 1.0371  data: 0.0224  max mem: 6508
Epoch: [7]  [ 50/104]  eta: 0:00:57  lr: 0.001000  loss: 0.3500 (0.3964)  loss_classifier: 0.1315 (0.1487)  loss_box_reg: 0.1885 (0.2147)  loss_objectness: 0.0163 (0.0185)  loss_rpn_box_reg: 0.0110 (0.0145)  time: 1.0241  data: 0.0218  max mem: 6508
Epoch: [7]  [ 60/104]  eta: 0:00:46  lr: 0.001000  loss: 0.3600 (0.3941)  loss_classifier: 0.1315 (0.1487)  loss_box_reg: 0.2080 (0.2129)  loss_objectness: 0.0157 (0.0183)  loss_rpn_box_reg: 0.0108 (0.0142)  time: 1.0117  data: 0.0209  max mem: 6508
Epoch: [7]  [ 70/104]  eta: 0:00:35  lr: 0.001000  loss: 0.3652 (0.3967)  loss_classifier: 0.1331 (0.1506)  loss_box_reg: 0.2129 (0.2147)  loss_objectness: 0.0148 (0.0177)  loss_rpn_box_reg: 0.0103 (0.0137)  time: 1.0107  data: 0.0219  max mem: 6508
Epoch: [7]  [ 80/104]  eta: 0:00:25  lr: 0.001000  loss: 0.3829 (0.3960)  loss_classifier: 0.1563 (0.1501)  loss_box_reg: 0.2107 (0.2148)  loss_objectness: 0.0132 (0.0172)  loss_rpn_box_reg: 0.0103 (0.0139)  time: 1.0236  data: 0.0241  max mem: 6508
Epoch: [7]  [ 90/104]  eta: 0:00:14  lr: 0.001000  loss: 0.4413 (0.4044)  loss_classifier: 0.1581 (0.1521)  loss_box_reg: 0.2467 (0.2207)  loss_objectness: 0.0141 (0.0171)  loss_rpn_box_reg: 0.0127 (0.0145)  time: 1.0289  data: 0.0229  max mem: 6508
Epoch: [7]  [100/104]  eta: 0:00:04  lr: 0.001000  loss: 0.4413 (0.4047)  loss_classifier: 0.1590 (0.1522)  loss_box_reg: 0.2542 (0.2211)  loss_objectness: 0.0126 (0.0169)  loss_rpn_box_reg: 0.0147 (0.0144)  time: 1.0227  data: 0.0199  max mem: 6508
Epoch: [7]  [103/104]  eta: 0:00:01  lr: 0.001000  loss: 0.4344 (0.4043)  loss_classifier: 0.1520 (0.1520)  loss_box_reg: 0.2392 (0.2209)  loss_objectness: 0.0126 (0.0171)  loss_rpn_box_reg: 0.0147 (0.0143)  time: 1.0264  data: 0.0201  max mem: 6508
Epoch: [7] Total time: 0:01:48 (1.0407 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:41  model_time: 0.4911 (0.4911)  evaluator_time: 0.0342 (0.0342)  time: 1.6021  data: 1.0465  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4356 (0.4383)  evaluator_time: 0.0215 (0.0262)  time: 0.4905  data: 0.0205  max mem: 6508
Test: Total time: 0:00:14 (0.5396 s / it)
Averaged stats: model_time: 0.4356 (0.4383)  evaluator_time: 0.0215 (0.0262)
Accumulating evaluation results...
DONE (t=0.33s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.355
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.700
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.315
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.288
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.406
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.244
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.157
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.442
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.506
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.434
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.532
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.317
Epoch: [8]  [  0/104]  eta: 0:04:34  lr: 0.001000  loss: 0.2473 (0.2473)  loss_classifier: 0.0932 (0.0932)  loss_box_reg: 0.1216 (0.1216)  loss_objectness: 0.0231 (0.0231)  loss_rpn_box_reg: 0.0095 (0.0095)  time: 2.6393  data: 1.2524  max mem: 6508
Epoch: [8]  [ 10/104]  eta: 0:01:52  lr: 0.001000  loss: 0.3371 (0.4016)  loss_classifier: 0.1366 (0.1563)  loss_box_reg: 0.1766 (0.2157)  loss_objectness: 0.0176 (0.0170)  loss_rpn_box_reg: 0.0103 (0.0127)  time: 1.2014  data: 0.1328  max mem: 6508
Epoch: [8]  [ 20/104]  eta: 0:01:35  lr: 0.001000  loss: 0.3752 (0.3958)  loss_classifier: 0.1365 (0.1530)  loss_box_reg: 0.1974 (0.2131)  loss_objectness: 0.0141 (0.0163)  loss_rpn_box_reg: 0.0125 (0.0134)  time: 1.0572  data: 0.0211  max mem: 6508
Epoch: [8]  [ 30/104]  eta: 0:01:22  lr: 0.001000  loss: 0.3752 (0.3949)  loss_classifier: 0.1365 (0.1489)  loss_box_reg: 0.2118 (0.2171)  loss_objectness: 0.0131 (0.0151)  loss_rpn_box_reg: 0.0138 (0.0137)  time: 1.0575  data: 0.0232  max mem: 6508
Epoch: [8]  [ 40/104]  eta: 0:01:09  lr: 0.001000  loss: 0.3710 (0.3867)  loss_classifier: 0.1305 (0.1442)  loss_box_reg: 0.2202 (0.2145)  loss_objectness: 0.0123 (0.0149)  loss_rpn_box_reg: 0.0119 (0.0131)  time: 1.0413  data: 0.0244  max mem: 6508
Epoch: [8]  [ 50/104]  eta: 0:00:57  lr: 0.001000  loss: 0.3710 (0.3918)  loss_classifier: 0.1323 (0.1467)  loss_box_reg: 0.2105 (0.2161)  loss_objectness: 0.0123 (0.0154)  loss_rpn_box_reg: 0.0106 (0.0136)  time: 1.0161  data: 0.0213  max mem: 6508
Epoch: [8]  [ 60/104]  eta: 0:00:46  lr: 0.001000  loss: 0.3706 (0.3911)  loss_classifier: 0.1415 (0.1471)  loss_box_reg: 0.1955 (0.2153)  loss_objectness: 0.0110 (0.0150)  loss_rpn_box_reg: 0.0138 (0.0138)  time: 1.0134  data: 0.0206  max mem: 6508
Epoch: [8]  [ 70/104]  eta: 0:00:35  lr: 0.001000  loss: 0.3706 (0.3889)  loss_classifier: 0.1418 (0.1461)  loss_box_reg: 0.1955 (0.2142)  loss_objectness: 0.0110 (0.0150)  loss_rpn_box_reg: 0.0138 (0.0136)  time: 1.0187  data: 0.0216  max mem: 6508
Epoch: [8]  [ 80/104]  eta: 0:00:25  lr: 0.001000  loss: 0.3642 (0.3899)  loss_classifier: 0.1418 (0.1457)  loss_box_reg: 0.2009 (0.2146)  loss_objectness: 0.0159 (0.0154)  loss_rpn_box_reg: 0.0133 (0.0141)  time: 1.0195  data: 0.0208  max mem: 6508
Epoch: [8]  [ 90/104]  eta: 0:00:14  lr: 0.001000  loss: 0.3939 (0.3970)  loss_classifier: 0.1465 (0.1479)  loss_box_reg: 0.2182 (0.2190)  loss_objectness: 0.0168 (0.0160)  loss_rpn_box_reg: 0.0131 (0.0141)  time: 1.0206  data: 0.0206  max mem: 6508
Epoch: [8]  [100/104]  eta: 0:00:04  lr: 0.001000  loss: 0.3779 (0.3900)  loss_classifier: 0.1397 (0.1455)  loss_box_reg: 0.2145 (0.2154)  loss_objectness: 0.0126 (0.0156)  loss_rpn_box_reg: 0.0105 (0.0136)  time: 1.0246  data: 0.0208  max mem: 6508
Epoch: [8]  [103/104]  eta: 0:00:01  lr: 0.001000  loss: 0.3779 (0.3899)  loss_classifier: 0.1397 (0.1455)  loss_box_reg: 0.2145 (0.2152)  loss_objectness: 0.0129 (0.0156)  loss_rpn_box_reg: 0.0105 (0.0136)  time: 1.0223  data: 0.0200  max mem: 6508
Epoch: [8] Total time: 0:01:48 (1.0475 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:40  model_time: 0.6768 (0.6768)  evaluator_time: 0.1020 (0.1020)  time: 1.5605  data: 0.7592  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4348 (0.4439)  evaluator_time: 0.0188 (0.0410)  time: 0.4799  data: 0.0191  max mem: 6508
Test: Total time: 0:00:14 (0.5454 s / it)
Averaged stats: model_time: 0.4348 (0.4439)  evaluator_time: 0.0188 (0.0410)
Accumulating evaluation results...
DONE (t=0.14s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.353
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.710
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.311
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.299
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.416
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.235
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.154
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.438
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.494
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.441
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.533
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.307
Epoch: [9]  [  0/104]  eta: 0:03:51  lr: 0.001000  loss: 0.2582 (0.2582)  loss_classifier: 0.0994 (0.0994)  loss_box_reg: 0.1455 (0.1455)  loss_objectness: 0.0069 (0.0069)  loss_rpn_box_reg: 0.0064 (0.0064)  time: 2.2229  data: 0.9032  max mem: 6508
Epoch: [9]  [ 10/104]  eta: 0:01:48  lr: 0.001000  loss: 0.4245 (0.3695)  loss_classifier: 0.1346 (0.1332)  loss_box_reg: 0.2167 (0.2120)  loss_objectness: 0.0120 (0.0134)  loss_rpn_box_reg: 0.0103 (0.0109)  time: 1.1574  data: 0.1001  max mem: 6508
Epoch: [9]  [ 20/104]  eta: 0:01:33  lr: 0.001000  loss: 0.4000 (0.3769)  loss_classifier: 0.1377 (0.1363)  loss_box_reg: 0.2237 (0.2162)  loss_objectness: 0.0132 (0.0133)  loss_rpn_box_reg: 0.0096 (0.0111)  time: 1.0624  data: 0.0226  max mem: 6508
Epoch: [9]  [ 30/104]  eta: 0:01:21  lr: 0.001000  loss: 0.3995 (0.3815)  loss_classifier: 0.1344 (0.1371)  loss_box_reg: 0.2237 (0.2193)  loss_objectness: 0.0121 (0.0129)  loss_rpn_box_reg: 0.0085 (0.0122)  time: 1.0685  data: 0.0258  max mem: 6508
Epoch: [9]  [ 40/104]  eta: 0:01:09  lr: 0.001000  loss: 0.4050 (0.3959)  loss_classifier: 0.1295 (0.1445)  loss_box_reg: 0.2357 (0.2252)  loss_objectness: 0.0120 (0.0130)  loss_rpn_box_reg: 0.0138 (0.0133)  time: 1.0399  data: 0.0230  max mem: 6508
Epoch: [9]  [ 50/104]  eta: 0:00:57  lr: 0.001000  loss: 0.3894 (0.3892)  loss_classifier: 0.1551 (0.1439)  loss_box_reg: 0.2282 (0.2193)  loss_objectness: 0.0118 (0.0129)  loss_rpn_box_reg: 0.0138 (0.0130)  time: 1.0168  data: 0.0216  max mem: 6508
Epoch: [9]  [ 60/104]  eta: 0:00:46  lr: 0.001000  loss: 0.3969 (0.3904)  loss_classifier: 0.1575 (0.1450)  loss_box_reg: 0.2105 (0.2176)  loss_objectness: 0.0138 (0.0142)  loss_rpn_box_reg: 0.0119 (0.0137)  time: 1.0163  data: 0.0238  max mem: 6508
Epoch: [9]  [ 70/104]  eta: 0:00:35  lr: 0.001000  loss: 0.4003 (0.3942)  loss_classifier: 0.1554 (0.1457)  loss_box_reg: 0.2174 (0.2193)  loss_objectness: 0.0166 (0.0145)  loss_rpn_box_reg: 0.0149 (0.0147)  time: 1.0169  data: 0.0235  max mem: 6508
Epoch: [9]  [ 80/104]  eta: 0:00:25  lr: 0.001000  loss: 0.3622 (0.3863)  loss_classifier: 0.1343 (0.1428)  loss_box_reg: 0.2125 (0.2151)  loss_objectness: 0.0145 (0.0143)  loss_rpn_box_reg: 0.0109 (0.0141)  time: 1.0167  data: 0.0216  max mem: 6508
Epoch: [9]  [ 90/104]  eta: 0:00:14  lr: 0.001000  loss: 0.3177 (0.3828)  loss_classifier: 0.1343 (0.1418)  loss_box_reg: 0.1839 (0.2130)  loss_objectness: 0.0109 (0.0142)  loss_rpn_box_reg: 0.0097 (0.0137)  time: 1.0233  data: 0.0214  max mem: 6508
Epoch: [9]  [100/104]  eta: 0:00:04  lr: 0.001000  loss: 0.3410 (0.3824)  loss_classifier: 0.1387 (0.1423)  loss_box_reg: 0.1886 (0.2123)  loss_objectness: 0.0117 (0.0144)  loss_rpn_box_reg: 0.0100 (0.0135)  time: 1.0315  data: 0.0208  max mem: 6508
Epoch: [9]  [103/104]  eta: 0:00:01  lr: 0.001000  loss: 0.3467 (0.3798)  loss_classifier: 0.1387 (0.1413)  loss_box_reg: 0.1878 (0.2109)  loss_objectness: 0.0140 (0.0144)  loss_rpn_box_reg: 0.0100 (0.0133)  time: 1.0285  data: 0.0197  max mem: 6508
Epoch: [9] Total time: 0:01:48 (1.0457 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:36  model_time: 0.4960 (0.4960)  evaluator_time: 0.0317 (0.0317)  time: 1.4057  data: 0.8676  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4394 (0.4397)  evaluator_time: 0.0209 (0.0260)  time: 0.4977  data: 0.0232  max mem: 6508
Test: Total time: 0:00:13 (0.5332 s / it)
Averaged stats: model_time: 0.4394 (0.4397)  evaluator_time: 0.0209 (0.0260)
Accumulating evaluation results...
DONE (t=0.15s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.361
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.709
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.307
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.323
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.422
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.232
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.159
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.449
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.509
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.447
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.544
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.322
Epoch: [10]  [  0/104]  eta: 0:03:24  lr: 0.001000  loss: 0.3358 (0.3358)  loss_classifier: 0.1290 (0.1290)  loss_box_reg: 0.1879 (0.1879)  loss_objectness: 0.0118 (0.0118)  loss_rpn_box_reg: 0.0071 (0.0071)  time: 1.9649  data: 0.8928  max mem: 6508
Epoch: [10]  [ 10/104]  eta: 0:01:47  lr: 0.001000  loss: 0.3452 (0.3785)  loss_classifier: 0.1273 (0.1465)  loss_box_reg: 0.1987 (0.2049)  loss_objectness: 0.0131 (0.0130)  loss_rpn_box_reg: 0.0083 (0.0141)  time: 1.1438  data: 0.1010  max mem: 6508
Epoch: [10]  [ 20/104]  eta: 0:01:32  lr: 0.001000  loss: 0.3529 (0.3655)  loss_classifier: 0.1237 (0.1324)  loss_box_reg: 0.2196 (0.2063)  loss_objectness: 0.0114 (0.0119)  loss_rpn_box_reg: 0.0111 (0.0150)  time: 1.0605  data: 0.0220  max mem: 6508
Epoch: [10]  [ 30/104]  eta: 0:01:20  lr: 0.001000  loss: 0.3529 (0.3614)  loss_classifier: 0.1161 (0.1324)  loss_box_reg: 0.2196 (0.2030)  loss_objectness: 0.0101 (0.0123)  loss_rpn_box_reg: 0.0108 (0.0136)  time: 1.0472  data: 0.0211  max mem: 6508
Epoch: [10]  [ 40/104]  eta: 0:01:08  lr: 0.001000  loss: 0.3148 (0.3574)  loss_classifier: 0.1141 (0.1295)  loss_box_reg: 0.1836 (0.2014)  loss_objectness: 0.0125 (0.0130)  loss_rpn_box_reg: 0.0104 (0.0135)  time: 1.0330  data: 0.0215  max mem: 6508
Epoch: [10]  [ 50/104]  eta: 0:00:57  lr: 0.001000  loss: 0.3396 (0.3581)  loss_classifier: 0.1231 (0.1299)  loss_box_reg: 0.2026 (0.2023)  loss_objectness: 0.0125 (0.0129)  loss_rpn_box_reg: 0.0113 (0.0131)  time: 1.0262  data: 0.0221  max mem: 6508
Epoch: [10]  [ 60/104]  eta: 0:00:46  lr: 0.001000  loss: 0.3396 (0.3611)  loss_classifier: 0.1231 (0.1306)  loss_box_reg: 0.1967 (0.2043)  loss_objectness: 0.0125 (0.0131)  loss_rpn_box_reg: 0.0107 (0.0132)  time: 1.0260  data: 0.0227  max mem: 6508
Epoch: [10]  [ 70/104]  eta: 0:00:35  lr: 0.001000  loss: 0.3545 (0.3665)  loss_classifier: 0.1419 (0.1340)  loss_box_reg: 0.2021 (0.2057)  loss_objectness: 0.0160 (0.0137)  loss_rpn_box_reg: 0.0101 (0.0131)  time: 1.0250  data: 0.0231  max mem: 6508
Epoch: [10]  [ 80/104]  eta: 0:00:25  lr: 0.001000  loss: 0.3728 (0.3656)  loss_classifier: 0.1448 (0.1329)  loss_box_reg: 0.2101 (0.2054)  loss_objectness: 0.0178 (0.0139)  loss_rpn_box_reg: 0.0101 (0.0134)  time: 1.0202  data: 0.0209  max mem: 6508
Epoch: [10]  [ 90/104]  eta: 0:00:14  lr: 0.001000  loss: 0.3848 (0.3708)  loss_classifier: 0.1362 (0.1356)  loss_box_reg: 0.2158 (0.2082)  loss_objectness: 0.0120 (0.0138)  loss_rpn_box_reg: 0.0096 (0.0132)  time: 1.0240  data: 0.0202  max mem: 6508
Epoch: [10]  [100/104]  eta: 0:00:04  lr: 0.001000  loss: 0.4097 (0.3745)  loss_classifier: 0.1414 (0.1367)  loss_box_reg: 0.2242 (0.2109)  loss_objectness: 0.0107 (0.0139)  loss_rpn_box_reg: 0.0098 (0.0130)  time: 1.0240  data: 0.0196  max mem: 6508
Epoch: [10]  [103/104]  eta: 0:00:01  lr: 0.001000  loss: 0.3852 (0.3717)  loss_classifier: 0.1362 (0.1359)  loss_box_reg: 0.2190 (0.2092)  loss_objectness: 0.0107 (0.0138)  loss_rpn_box_reg: 0.0092 (0.0129)  time: 1.0259  data: 0.0195  max mem: 6508
Epoch: [10] Total time: 0:01:48 (1.0431 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:35  model_time: 0.5398 (0.5398)  evaluator_time: 0.0720 (0.0720)  time: 1.3554  data: 0.7296  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4400 (0.4421)  evaluator_time: 0.0213 (0.0269)  time: 0.4923  data: 0.0208  max mem: 6508
Test: Total time: 0:00:13 (0.5286 s / it)
Averaged stats: model_time: 0.4400 (0.4421)  evaluator_time: 0.0213 (0.0269)
Accumulating evaluation results...
DONE (t=0.15s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.366
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.724
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.323
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.310
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.423
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.215
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.153
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.447
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.509
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.455
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.544
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.289
Epoch: [11]  [  0/104]  eta: 0:03:21  lr: 0.001000  loss: 0.3552 (0.3552)  loss_classifier: 0.1059 (0.1059)  loss_box_reg: 0.2213 (0.2213)  loss_objectness: 0.0111 (0.0111)  loss_rpn_box_reg: 0.0169 (0.0169)  time: 1.9349  data: 0.8788  max mem: 6508
Epoch: [11]  [ 10/104]  eta: 0:01:46  lr: 0.001000  loss: 0.3552 (0.3607)  loss_classifier: 0.1188 (0.1305)  loss_box_reg: 0.2090 (0.2034)  loss_objectness: 0.0128 (0.0137)  loss_rpn_box_reg: 0.0129 (0.0130)  time: 1.1327  data: 0.0965  max mem: 6508
Epoch: [11]  [ 20/104]  eta: 0:01:31  lr: 0.001000  loss: 0.3827 (0.3783)  loss_classifier: 0.1198 (0.1367)  loss_box_reg: 0.2090 (0.2148)  loss_objectness: 0.0125 (0.0143)  loss_rpn_box_reg: 0.0116 (0.0125)  time: 1.0513  data: 0.0190  max mem: 6508
Epoch: [11]  [ 30/104]  eta: 0:01:19  lr: 0.001000  loss: 0.3661 (0.3711)  loss_classifier: 0.1198 (0.1348)  loss_box_reg: 0.2020 (0.2102)  loss_objectness: 0.0122 (0.0132)  loss_rpn_box_reg: 0.0103 (0.0128)  time: 1.0449  data: 0.0216  max mem: 6508
Epoch: [11]  [ 40/104]  eta: 0:01:08  lr: 0.001000  loss: 0.2960 (0.3559)  loss_classifier: 0.1169 (0.1307)  loss_box_reg: 0.1678 (0.2006)  loss_objectness: 0.0103 (0.0131)  loss_rpn_box_reg: 0.0082 (0.0114)  time: 1.0306  data: 0.0219  max mem: 6508
Epoch: [11]  [ 50/104]  eta: 0:00:56  lr: 0.001000  loss: 0.2937 (0.3545)  loss_classifier: 0.1132 (0.1304)  loss_box_reg: 0.1621 (0.1999)  loss_objectness: 0.0102 (0.0129)  loss_rpn_box_reg: 0.0082 (0.0113)  time: 1.0160  data: 0.0204  max mem: 6508
Epoch: [11]  [ 60/104]  eta: 0:00:45  lr: 0.001000  loss: 0.3493 (0.3636)  loss_classifier: 0.1264 (0.1330)  loss_box_reg: 0.1996 (0.2051)  loss_objectness: 0.0117 (0.0131)  loss_rpn_box_reg: 0.0116 (0.0124)  time: 1.0080  data: 0.0202  max mem: 6508
Epoch: [11]  [ 70/104]  eta: 0:00:35  lr: 0.001000  loss: 0.3441 (0.3586)  loss_classifier: 0.1179 (0.1313)  loss_box_reg: 0.1896 (0.2018)  loss_objectness: 0.0141 (0.0135)  loss_rpn_box_reg: 0.0105 (0.0119)  time: 1.0125  data: 0.0210  max mem: 6508
Epoch: [11]  [ 80/104]  eta: 0:00:24  lr: 0.001000  loss: 0.3341 (0.3598)  loss_classifier: 0.1228 (0.1318)  loss_box_reg: 0.1810 (0.2028)  loss_objectness: 0.0119 (0.0132)  loss_rpn_box_reg: 0.0105 (0.0121)  time: 1.0234  data: 0.0224  max mem: 6508
Epoch: [11]  [ 90/104]  eta: 0:00:14  lr: 0.001000  loss: 0.3917 (0.3616)  loss_classifier: 0.1369 (0.1324)  loss_box_reg: 0.2154 (0.2037)  loss_objectness: 0.0131 (0.0133)  loss_rpn_box_reg: 0.0124 (0.0123)  time: 1.0307  data: 0.0232  max mem: 6508
Epoch: [11]  [100/104]  eta: 0:00:04  lr: 0.001000  loss: 0.3986 (0.3644)  loss_classifier: 0.1348 (0.1332)  loss_box_reg: 0.2154 (0.2056)  loss_objectness: 0.0115 (0.0131)  loss_rpn_box_reg: 0.0124 (0.0125)  time: 1.0378  data: 0.0231  max mem: 6508
Epoch: [11]  [103/104]  eta: 0:00:01  lr: 0.001000  loss: 0.3917 (0.3665)  loss_classifier: 0.1388 (0.1337)  loss_box_reg: 0.2154 (0.2069)  loss_objectness: 0.0113 (0.0131)  loss_rpn_box_reg: 0.0134 (0.0128)  time: 1.0384  data: 0.0229  max mem: 6508
Epoch: [11] Total time: 0:01:48 (1.0407 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:45  model_time: 0.5142 (0.5142)  evaluator_time: 0.1822 (0.1822)  time: 1.7422  data: 1.0378  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4337 (0.4395)  evaluator_time: 0.0174 (0.0288)  time: 0.4793  data: 0.0206  max mem: 6508
Test: Total time: 0:00:14 (0.5425 s / it)
Averaged stats: model_time: 0.4337 (0.4395)  evaluator_time: 0.0174 (0.0288)
Accumulating evaluation results...
DONE (t=0.15s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.371
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.732
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.336
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.301
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.429
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.254
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.161
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.451
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.513
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.445
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.550
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.339
Epoch: [12]  [  0/104]  eta: 0:04:01  lr: 0.001000  loss: 0.3296 (0.3296)  loss_classifier: 0.1243 (0.1243)  loss_box_reg: 0.1716 (0.1716)  loss_objectness: 0.0247 (0.0247)  loss_rpn_box_reg: 0.0090 (0.0090)  time: 2.3198  data: 0.9138  max mem: 6508
Epoch: [12]  [ 10/104]  eta: 0:01:49  lr: 0.001000  loss: 0.3754 (0.4018)  loss_classifier: 0.1278 (0.1475)  loss_box_reg: 0.2398 (0.2275)  loss_objectness: 0.0135 (0.0140)  loss_rpn_box_reg: 0.0115 (0.0128)  time: 1.1628  data: 0.1006  max mem: 6508
Epoch: [12]  [ 20/104]  eta: 0:01:33  lr: 0.001000  loss: 0.3603 (0.3675)  loss_classifier: 0.1220 (0.1319)  loss_box_reg: 0.2210 (0.2105)  loss_objectness: 0.0120 (0.0124)  loss_rpn_box_reg: 0.0115 (0.0126)  time: 1.0503  data: 0.0203  max mem: 6508
Epoch: [12]  [ 30/104]  eta: 0:01:20  lr: 0.001000  loss: 0.3440 (0.3694)  loss_classifier: 0.1220 (0.1362)  loss_box_reg: 0.1879 (0.2081)  loss_objectness: 0.0107 (0.0127)  loss_rpn_box_reg: 0.0107 (0.0124)  time: 1.0448  data: 0.0215  max mem: 6508
Epoch: [12]  [ 40/104]  eta: 0:01:08  lr: 0.001000  loss: 0.3340 (0.3627)  loss_classifier: 0.1333 (0.1333)  loss_box_reg: 0.1853 (0.2040)  loss_objectness: 0.0133 (0.0129)  loss_rpn_box_reg: 0.0101 (0.0125)  time: 1.0277  data: 0.0216  max mem: 6508
Epoch: [12]  [ 50/104]  eta: 0:00:57  lr: 0.001000  loss: 0.3332 (0.3669)  loss_classifier: 0.1333 (0.1340)  loss_box_reg: 0.1968 (0.2074)  loss_objectness: 0.0113 (0.0127)  loss_rpn_box_reg: 0.0101 (0.0129)  time: 1.0174  data: 0.0218  max mem: 6508
Epoch: [12]  [ 60/104]  eta: 0:00:46  lr: 0.001000  loss: 0.3676 (0.3705)  loss_classifier: 0.1290 (0.1345)  loss_box_reg: 0.2198 (0.2096)  loss_objectness: 0.0129 (0.0132)  loss_rpn_box_reg: 0.0118 (0.0131)  time: 1.0239  data: 0.0242  max mem: 6508
Epoch: [12]  [ 70/104]  eta: 0:00:35  lr: 0.001000  loss: 0.3456 (0.3656)  loss_classifier: 0.1198 (0.1329)  loss_box_reg: 0.1968 (0.2069)  loss_objectness: 0.0132 (0.0132)  loss_rpn_box_reg: 0.0099 (0.0126)  time: 1.0260  data: 0.0231  max mem: 6508
Epoch: [12]  [ 80/104]  eta: 0:00:25  lr: 0.001000  loss: 0.3438 (0.3698)  loss_classifier: 0.1352 (0.1343)  loss_box_reg: 0.1954 (0.2089)  loss_objectness: 0.0120 (0.0134)  loss_rpn_box_reg: 0.0099 (0.0132)  time: 1.0216  data: 0.0201  max mem: 6508
Epoch: [12]  [ 90/104]  eta: 0:00:14  lr: 0.001000  loss: 0.3364 (0.3691)  loss_classifier: 0.1323 (0.1338)  loss_box_reg: 0.1908 (0.2090)  loss_objectness: 0.0100 (0.0131)  loss_rpn_box_reg: 0.0101 (0.0132)  time: 1.0283  data: 0.0202  max mem: 6508
Epoch: [12]  [100/104]  eta: 0:00:04  lr: 0.001000  loss: 0.3178 (0.3666)  loss_classifier: 0.1175 (0.1327)  loss_box_reg: 0.1908 (0.2086)  loss_objectness: 0.0087 (0.0126)  loss_rpn_box_reg: 0.0082 (0.0128)  time: 1.0286  data: 0.0214  max mem: 6508
Epoch: [12]  [103/104]  eta: 0:00:01  lr: 0.001000  loss: 0.3118 (0.3633)  loss_classifier: 0.1122 (0.1314)  loss_box_reg: 0.1895 (0.2065)  loss_objectness: 0.0087 (0.0127)  loss_rpn_box_reg: 0.0085 (0.0127)  time: 1.0235  data: 0.0212  max mem: 6508
Epoch: [12] Total time: 0:01:48 (1.0437 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:41  model_time: 0.5193 (0.5193)  evaluator_time: 0.3265 (0.3265)  time: 1.6055  data: 0.7408  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4387 (0.4419)  evaluator_time: 0.0244 (0.0428)  time: 0.5053  data: 0.0242  max mem: 6508
Test: Total time: 0:00:14 (0.5500 s / it)
Averaged stats: model_time: 0.4387 (0.4419)  evaluator_time: 0.0244 (0.0428)
Accumulating evaluation results...
DONE (t=0.15s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.372
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.738
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.323
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.306
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.427
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.267
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.159
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.454
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.513
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.441
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.547
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.368
Epoch: [13]  [  0/104]  eta: 0:03:34  lr: 0.001000  loss: 0.3092 (0.3092)  loss_classifier: 0.0881 (0.0881)  loss_box_reg: 0.1915 (0.1915)  loss_objectness: 0.0168 (0.0168)  loss_rpn_box_reg: 0.0128 (0.0128)  time: 2.0664  data: 1.0264  max mem: 6508
Epoch: [13]  [ 10/104]  eta: 0:01:46  lr: 0.001000  loss: 0.3466 (0.3489)  loss_classifier: 0.1134 (0.1258)  loss_box_reg: 0.1997 (0.1976)  loss_objectness: 0.0141 (0.0151)  loss_rpn_box_reg: 0.0102 (0.0104)  time: 1.1362  data: 0.1088  max mem: 6508
Epoch: [13]  [ 20/104]  eta: 0:01:33  lr: 0.001000  loss: 0.3585 (0.3608)  loss_classifier: 0.1336 (0.1343)  loss_box_reg: 0.1997 (0.2015)  loss_objectness: 0.0132 (0.0142)  loss_rpn_box_reg: 0.0102 (0.0108)  time: 1.0621  data: 0.0195  max mem: 6508
Epoch: [13]  [ 30/104]  eta: 0:01:20  lr: 0.001000  loss: 0.3632 (0.3590)  loss_classifier: 0.1322 (0.1303)  loss_box_reg: 0.2048 (0.2051)  loss_objectness: 0.0106 (0.0130)  loss_rpn_box_reg: 0.0103 (0.0106)  time: 1.0673  data: 0.0204  max mem: 6508
Epoch: [13]  [ 40/104]  eta: 0:01:08  lr: 0.001000  loss: 0.3632 (0.3617)  loss_classifier: 0.1311 (0.1330)  loss_box_reg: 0.2053 (0.2040)  loss_objectness: 0.0106 (0.0131)  loss_rpn_box_reg: 0.0099 (0.0116)  time: 1.0441  data: 0.0203  max mem: 6508
Epoch: [13]  [ 50/104]  eta: 0:00:57  lr: 0.001000  loss: 0.3369 (0.3605)  loss_classifier: 0.1256 (0.1294)  loss_box_reg: 0.1936 (0.2067)  loss_objectness: 0.0104 (0.0126)  loss_rpn_box_reg: 0.0092 (0.0118)  time: 1.0233  data: 0.0217  max mem: 6508
Epoch: [13]  [ 60/104]  eta: 0:00:46  lr: 0.001000  loss: 0.3831 (0.3630)  loss_classifier: 0.1317 (0.1311)  loss_box_reg: 0.2342 (0.2079)  loss_objectness: 0.0078 (0.0118)  loss_rpn_box_reg: 0.0096 (0.0121)  time: 1.0105  data: 0.0221  max mem: 6508
Epoch: [13]  [ 70/104]  eta: 0:00:35  lr: 0.001000  loss: 0.3831 (0.3632)  loss_classifier: 0.1336 (0.1317)  loss_box_reg: 0.2201 (0.2072)  loss_objectness: 0.0090 (0.0120)  loss_rpn_box_reg: 0.0122 (0.0124)  time: 1.0019  data: 0.0209  max mem: 6508
Epoch: [13]  [ 80/104]  eta: 0:00:25  lr: 0.001000  loss: 0.3283 (0.3581)  loss_classifier: 0.1210 (0.1308)  loss_box_reg: 0.1896 (0.2034)  loss_objectness: 0.0094 (0.0119)  loss_rpn_box_reg: 0.0101 (0.0120)  time: 1.0087  data: 0.0218  max mem: 6508
Epoch: [13]  [ 90/104]  eta: 0:00:14  lr: 0.001000  loss: 0.3066 (0.3566)  loss_classifier: 0.1091 (0.1300)  loss_box_reg: 0.1808 (0.2029)  loss_objectness: 0.0094 (0.0117)  loss_rpn_box_reg: 0.0087 (0.0120)  time: 1.0290  data: 0.0224  max mem: 6508
Epoch: [13]  [100/104]  eta: 0:00:04  lr: 0.001000  loss: 0.3514 (0.3582)  loss_classifier: 0.1135 (0.1301)  loss_box_reg: 0.2078 (0.2040)  loss_objectness: 0.0110 (0.0119)  loss_rpn_box_reg: 0.0090 (0.0123)  time: 1.0397  data: 0.0198  max mem: 6508
Epoch: [13]  [103/104]  eta: 0:00:01  lr: 0.001000  loss: 0.3514 (0.3583)  loss_classifier: 0.1213 (0.1303)  loss_box_reg: 0.2078 (0.2038)  loss_objectness: 0.0110 (0.0118)  loss_rpn_box_reg: 0.0119 (0.0124)  time: 1.0423  data: 0.0195  max mem: 6508
Epoch: [13] Total time: 0:01:48 (1.0442 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:35  model_time: 0.5151 (0.5151)  evaluator_time: 0.0298 (0.0298)  time: 1.3750  data: 0.8028  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4341 (0.4373)  evaluator_time: 0.0164 (0.0217)  time: 0.4862  data: 0.0221  max mem: 6508
Test: Total time: 0:00:13 (0.5220 s / it)
Averaged stats: model_time: 0.4341 (0.4373)  evaluator_time: 0.0164 (0.0217)
Accumulating evaluation results...
DONE (t=0.15s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.380
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.727
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.353
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.298
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.424
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.257
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.166
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.455
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.517
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.450
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.552
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.362
Epoch: [14]  [  0/104]  eta: 0:02:58  lr: 0.001000  loss: 0.4111 (0.4111)  loss_classifier: 0.1705 (0.1705)  loss_box_reg: 0.2171 (0.2171)  loss_objectness: 0.0078 (0.0078)  loss_rpn_box_reg: 0.0156 (0.0156)  time: 1.7151  data: 0.5224  max mem: 6508
Epoch: [14]  [ 10/104]  eta: 0:01:43  lr: 0.001000  loss: 0.3929 (0.3500)  loss_classifier: 0.1241 (0.1285)  loss_box_reg: 0.2171 (0.1973)  loss_objectness: 0.0107 (0.0108)  loss_rpn_box_reg: 0.0111 (0.0133)  time: 1.1013  data: 0.0654  max mem: 6508
Epoch: [14]  [ 20/104]  eta: 0:01:30  lr: 0.001000  loss: 0.3124 (0.3298)  loss_classifier: 0.1204 (0.1192)  loss_box_reg: 0.1759 (0.1870)  loss_objectness: 0.0101 (0.0112)  loss_rpn_box_reg: 0.0100 (0.0124)  time: 1.0395  data: 0.0203  max mem: 6508
Epoch: [14]  [ 30/104]  eta: 0:01:18  lr: 0.001000  loss: 0.3136 (0.3314)  loss_classifier: 0.1137 (0.1202)  loss_box_reg: 0.1806 (0.1869)  loss_objectness: 0.0101 (0.0113)  loss_rpn_box_reg: 0.0090 (0.0130)  time: 1.0452  data: 0.0235  max mem: 6508
Epoch: [14]  [ 40/104]  eta: 0:01:07  lr: 0.001000  loss: 0.3399 (0.3387)  loss_classifier: 0.1130 (0.1212)  loss_box_reg: 0.1890 (0.1926)  loss_objectness: 0.0109 (0.0118)  loss_rpn_box_reg: 0.0090 (0.0131)  time: 1.0414  data: 0.0244  max mem: 6508
Epoch: [14]  [ 50/104]  eta: 0:00:56  lr: 0.001000  loss: 0.3399 (0.3483)  loss_classifier: 0.1125 (0.1252)  loss_box_reg: 0.2136 (0.1978)  loss_objectness: 0.0117 (0.0120)  loss_rpn_box_reg: 0.0099 (0.0133)  time: 1.0269  data: 0.0220  max mem: 6508
Epoch: [14]  [ 60/104]  eta: 0:00:45  lr: 0.001000  loss: 0.3414 (0.3437)  loss_classifier: 0.1187 (0.1243)  loss_box_reg: 0.1988 (0.1951)  loss_objectness: 0.0109 (0.0117)  loss_rpn_box_reg: 0.0099 (0.0127)  time: 1.0209  data: 0.0205  max mem: 6508
Epoch: [14]  [ 70/104]  eta: 0:00:35  lr: 0.001000  loss: 0.3355 (0.3447)  loss_classifier: 0.1128 (0.1246)  loss_box_reg: 0.1946 (0.1955)  loss_objectness: 0.0098 (0.0121)  loss_rpn_box_reg: 0.0096 (0.0124)  time: 1.0244  data: 0.0207  max mem: 6508
Epoch: [14]  [ 80/104]  eta: 0:00:24  lr: 0.001000  loss: 0.3644 (0.3475)  loss_classifier: 0.1185 (0.1261)  loss_box_reg: 0.1995 (0.1973)  loss_objectness: 0.0101 (0.0118)  loss_rpn_box_reg: 0.0096 (0.0123)  time: 1.0287  data: 0.0222  max mem: 6508
Epoch: [14]  [ 90/104]  eta: 0:00:14  lr: 0.001000  loss: 0.3639 (0.3469)  loss_classifier: 0.1278 (0.1253)  loss_box_reg: 0.2112 (0.1976)  loss_objectness: 0.0102 (0.0118)  loss_rpn_box_reg: 0.0093 (0.0122)  time: 1.0274  data: 0.0234  max mem: 6508
Epoch: [14]  [100/104]  eta: 0:00:04  lr: 0.001000  loss: 0.3578 (0.3490)  loss_classifier: 0.1257 (0.1258)  loss_box_reg: 0.2131 (0.1993)  loss_objectness: 0.0088 (0.0116)  loss_rpn_box_reg: 0.0121 (0.0123)  time: 1.0202  data: 0.0212  max mem: 6508
Epoch: [14]  [103/104]  eta: 0:00:01  lr: 0.001000  loss: 0.3658 (0.3506)  loss_classifier: 0.1278 (0.1266)  loss_box_reg: 0.2145 (0.2002)  loss_objectness: 0.0079 (0.0114)  loss_rpn_box_reg: 0.0115 (0.0123)  time: 1.0212  data: 0.0212  max mem: 6508
Epoch: [14] Total time: 0:01:48 (1.0386 s / it)
creating index...
index created!
Test:  [ 0/26]  eta: 0:00:50  model_time: 0.6277 (0.6277)  evaluator_time: 0.1529 (0.1529)  time: 1.9418  data: 1.1288  max mem: 6508
Test:  [25/26]  eta: 0:00:00  model_time: 0.4307 (0.4378)  evaluator_time: 0.0172 (0.0249)  time: 0.4760  data: 0.0193  max mem: 6508
Test: Total time: 0:00:14 (0.5392 s / it)
Averaged stats: model_time: 0.4307 (0.4378)  evaluator_time: 0.0172 (0.0249)
Accumulating evaluation results...
DONE (t=0.23s).
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.391
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.762
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.363
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.314
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.436
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.260
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.165
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.464
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.529
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.463
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.569
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.375

import re

# Define dictionaries to hold your data
metrics = {
    "AR_1": [],
    "AR_10": [],
    "AR_100": [],
    "AR_small": [],
    "AR_medium": [],
    "AR_large": [],
    "AP": [],           # Add AP metric
    "AP_50": [],        # Add AP_50 metric
    "AP_75": [],        # Add AP_75 metric
    "loss": [],         # Add loss metric
    "loss_classifier": [],  # Add loss_classifier metric
    "loss_box_reg": [],     # Add loss_box_reg metric
    "loss_objectness": [],  # Add loss_objectness metric
    "loss_rpn_box_reg": [],  # Add loss_rpn_box_reg metric
    "model_time": [],        # Add model_time metric
    "evaluator_time": [],    # Add evaluator_time metric
    "total_time": []         # Add total_time metric
}

# Regex patterns to extract data
pattern_ar_1 = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area=   all \| maxDets=  1 \] = (\d+\.\d+)")
pattern_ar_10 = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area=   all \| maxDets= 10 \] = (\d+\.\d+)")
pattern_ar_100 = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area=   all \| maxDets=100 \] = (\d+\.\d+)")
pattern_ar_small = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area= small \| maxDets=100 \] = (\d+\.\d+)")
pattern_ar_medium = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area=medium \| maxDets=100 \] = (\d+\.\d+)")
pattern_ar_large = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area= large \| maxDets=100 \] = (\d+\.\d+)")
pattern_ap = re.compile(r"Average Precision\s+\(AP\) @\[ IoU=0\.50:0\.95 \| area=   all \| maxDets=100 \] = (\d+\.\d+)")
pattern_ap_50 = re.compile(r"Average Precision\s+\(AP\) @\[ IoU=0\.50 \| area=   all \| maxDets=100 \] = (\d+\.\d+)")
pattern_ap_75 = re.compile(r"Average Precision\s+\(AP\) @\[ IoU=0\.75 \| area=   all \| maxDets=100 \] = (\d+\.\d+)")
pattern_loss = re.compile(r"loss: (\d+\.\d+)")                  # Pattern for loss
pattern_loss_classifier = re.compile(r"loss_classifier: (\d+\.\d+)")  # Pattern for loss_classifier
pattern_loss_box_reg = re.compile(r"loss_box_reg: (\d+\.\d+)")        # Pattern for loss_box_reg
pattern_loss_objectness = re.compile(r"loss_objectness: (\d+\.\d+)")  # Pattern for loss_objectness
pattern_loss_rpn_box_reg = re.compile(r"loss_rpn_box_reg: (\d+\.\d+)")# Pattern for loss_rpn_box_reg
pattern_model_time = re.compile(r"Model time:\s+(\d+\.\d+)")        # Pattern for model_time
pattern_evaluator_time = re.compile(r"Evaluator time:\s+(\d+\.\d+)")# Pattern for evaluator_time
pattern_total_time = re.compile(r"Total time:\s+(\d+\.\d+)")        # Pattern for total_time

# Read the log file
with open('eva adamn_sgd.txt', 'r') as file:
    file_content = file.read()

    # Handling AR matches
    metrics["AR_1"].extend([float(x) for x in pattern_ar_1.findall(file_content)])
    metrics["AR_10"].extend([float(x) for x in pattern_ar_10.findall(file_content)])
    metrics["AR_100"].extend([float(x) for x in pattern_ar_100.findall(file_content)])
    metrics["AR_small"].extend([float(x) for x in pattern_ar_small.findall(file_content)])
    metrics["AR_medium"].extend([float(x) for x in pattern_ar_medium.findall(file_content)])
    metrics["AR_large"].extend([float(x) for x in pattern_ar_large.findall(file_content)])

    # Handling AP matches
    metrics["AP"].extend([float(x) for x in pattern_ap.findall(file_content)])
    metrics["AP_50"].extend([float(x) for x in pattern_ap_50.findall(file_content)])
    metrics["AP_75"].extend([float(x) for x in pattern_ap_75.findall(file_content)])

    # Handling loss matches
    metrics["loss"].extend([float(x) for x in pattern_loss.findall(file_content)])

    # Handling loss_classifier matches
    metrics["loss_classifier"].extend([float(x) for x in pattern_loss_classifier.findall(file_content)])

    # Handling loss_box_reg matches
    metrics["loss_box_reg"].extend([float(x) for x in pattern_loss_box_reg.findall(file_content)])

    # Handling loss_objectness matches
    metrics["loss_objectness"].extend([float(x) for x in pattern_loss_objectness.findall(file_content)])

    # Handling loss_rpn_box_reg matches
    metrics["loss_rpn_box_reg"].extend([float(x) for x in pattern_loss_rpn_box_reg.findall(file_content)])

    # Handling model_time matches
    metrics["model_time"].extend([float(x) for x in pattern_model_time.findall(file_content)])

    # Handling evaluator_time matches
    metrics["evaluator_time"].extend([float(x) for x in pattern_evaluator_time.findall(file_content)])

    # Handling total_time matches
    metrics["total_time"].extend([float(x) for x in pattern_total_time.findall(file_content)])

# Print the collected metrics to verify
for key, value in metrics.items():
    print(f"{key}: {value}")

AR_1: [0.073, 0.007, 0.045, 0.071, 0.105, 0.123, 0.153, 0.157, 0.154, 0.159, 0.153, 0.161, 0.159, 0.166, 0.165]
AR_10: [0.178, 0.056, 0.174, 0.259, 0.315, 0.375, 0.433, 0.442, 0.438, 0.449, 0.447, 0.451, 0.454, 0.455, 0.464]
AR_100: [0.223, 0.101, 0.226, 0.323, 0.371, 0.441, 0.501, 0.506, 0.494, 0.509, 0.509, 0.513, 0.513, 0.517, 0.529]
AR_small: [0.246, 0.115, 0.235, 0.334, 0.35, 0.376, 0.413, 0.434, 0.441, 0.447, 0.455, 0.445, 0.441, 0.45, 0.463]
AR_medium: [0.178, 0.154, 0.278, 0.374, 0.373, 0.504, 0.553, 0.532, 0.533, 0.544, 0.544, 0.55, 0.547, 0.552, 0.569]
AR_large: [0.221, 0.0, 0.09, 0.06, 0.159, 0.237, 0.329, 0.317, 0.307, 0.322, 0.289, 0.339, 0.368, 0.362, 0.375]
AP: [0.098, 0.032, 0.095, 0.173, 0.225, 0.29, 0.347, 0.355, 0.353, 0.361, 0.366, 0.371, 0.372, 0.38, 0.391]
AP_50: []
AP_75: []
loss: [3.3583, 2.3279, 1.209, 0.9238, 0.879, 0.941, 0.8082, 0.6632, 0.5934, 0.8766, 0.7816, 0.6924, 0.5595, 0.9238, 0.9238, 0.8937, 0.8828, 0.8035, 0.7846, 0.7665, 0.7328, 0.8004, 0.8303, 0.8303, 0.5037, 0.7679, 0.7679, 0.6997, 0.6524, 0.6844, 0.6844, 0.7528, 0.8539, 0.8539, 0.6873, 0.6611, 0.9499, 0.8302, 0.6848, 0.7425, 0.7538, 0.6383, 0.6383, 0.5951, 0.6758, 0.6085, 0.5904, 0.6085, 0.9558, 0.7101, 0.6309, 0.6323, 0.6639, 0.575, 0.633, 0.6562, 0.6114, 0.5727, 0.5472, 0.5416, 0.468, 0.4746, 0.5318, 0.6038, 0.581, 0.5266, 0.5092, 0.4859, 0.5286, 0.5525, 0.5126, 0.4816, 0.2004, 0.379, 0.379, 0.3963, 0.4557, 0.4646, 0.4693, 0.4249, 0.4432, 0.4432, 0.3908, 0.3908, 0.5793, 0.3872, 0.3806, 0.4133, 0.3974, 0.35, 0.36, 0.3652, 0.3829, 0.4413, 0.4413, 0.4344, 0.2473, 0.3371, 0.3752, 0.3752, 0.371, 0.371, 0.3706, 0.3706, 0.3642, 0.3939, 0.3779, 0.3779, 0.2582, 0.4245, 0.4, 0.3995, 0.405, 0.3894, 0.3969, 0.4003, 0.3622, 0.3177, 0.341, 0.3467, 0.3358, 0.3452, 0.3529, 0.3529, 0.3148, 0.3396, 0.3396, 0.3545, 0.3728, 0.3848, 0.4097, 0.3852, 0.3552, 0.3552, 0.3827, 0.3661, 0.296, 0.2937, 0.3493, 0.3441, 0.3341, 0.3917, 0.3986, 0.3917, 0.3296, 0.3754, 0.3603, 0.344, 0.334, 0.3332, 0.3676, 0.3456, 0.3438, 0.3364, 0.3178, 0.3118, 0.3092, 0.3466, 0.3585, 0.3632, 0.3632, 0.3369, 0.3831, 0.3831, 0.3283, 0.3066, 0.3514, 0.3514, 0.4111, 0.3929, 0.3124, 0.3136, 0.3399, 0.3399, 0.3414, 0.3355, 0.3644, 0.3639, 0.3578, 0.3658]
loss_classifier: [2.5736, 1.5769, 0.664, 0.4708, 0.4517, 0.4517, 0.3806, 0.3152, 0.3131, 0.3554, 0.3472, 0.3055, 0.2305, 0.3809, 0.4395, 0.4452, 0.4144, 0.3222, 0.3764, 0.3365, 0.3365, 0.3582, 0.3423, 0.3094, 0.2505, 0.263, 0.3391, 0.3233, 0.3024, 0.3101, 0.3198, 0.3344, 0.3837, 0.3838, 0.2968, 0.2968, 0.3417, 0.3417, 0.3111, 0.3458, 0.282, 0.2502, 0.2553, 0.2605, 0.2742, 0.2477, 0.2427, 0.2726, 0.493, 0.2657, 0.2392, 0.2409, 0.275, 0.2283, 0.2227, 0.2589, 0.2353, 0.2379, 0.2376, 0.2112, 0.1478, 0.2071, 0.2175, 0.235, 0.235, 0.2094, 0.1943, 0.1813, 0.1831, 0.1898, 0.1921, 0.1829, 0.085, 0.1573, 0.1573, 0.1639, 0.1723, 0.1774, 0.1746, 0.1655, 0.1613, 0.1654, 0.1517, 0.1499, 0.2718, 0.1535, 0.142, 0.1405, 0.1418, 0.1315, 0.1315, 0.1331, 0.1563, 0.1581, 0.159, 0.152, 0.0932, 0.1366, 0.1365, 0.1365, 0.1305, 0.1323, 0.1415, 0.1418, 0.1418, 0.1465, 0.1397, 0.1397, 0.0994, 0.1346, 0.1377, 0.1344, 0.1295, 0.1551, 0.1575, 0.1554, 0.1343, 0.1343, 0.1387, 0.1387, 0.129, 0.1273, 0.1237, 0.1161, 0.1141, 0.1231, 0.1231, 0.1419, 0.1448, 0.1362, 0.1414, 0.1362, 0.1059, 0.1188, 0.1198, 0.1198, 0.1169, 0.1132, 0.1264, 0.1179, 0.1228, 0.1369, 0.1348, 0.1388, 0.1243, 0.1278, 0.122, 0.122, 0.1333, 0.1333, 0.129, 0.1198, 0.1352, 0.1323, 0.1175, 0.1122, 0.0881, 0.1134, 0.1336, 0.1322, 0.1311, 0.1256, 0.1317, 0.1336, 0.121, 0.1091, 0.1135, 0.1213, 0.1705, 0.1241, 0.1204, 0.1137, 0.113, 0.1125, 0.1187, 0.1128, 0.1185, 0.1278, 0.1257, 0.1278]
loss_box_reg: [0.4333, 0.359, 0.359, 0.3461, 0.3577, 0.3279, 0.3105, 0.2753, 0.2408, 0.3787, 0.3829, 0.3446, 0.2976, 0.42, 0.3396, 0.257, 0.2087, 0.1765, 0.2107, 0.178, 0.207, 0.2739, 0.2363, 0.2845, 0.1813, 0.3321, 0.3321, 0.2844, 0.238, 0.2446, 0.2572, 0.3064, 0.3581, 0.3581, 0.2919, 0.2749, 0.4621, 0.3616, 0.3178, 0.3276, 0.3276, 0.3137, 0.2984, 0.2664, 0.3144, 0.3144, 0.2899, 0.31, 0.3486, 0.3435, 0.293, 0.297, 0.297, 0.2813, 0.2909, 0.3541, 0.2933, 0.2812, 0.2812, 0.2812, 0.2861, 0.2416, 0.2739, 0.289, 0.2642, 0.2642, 0.2546, 0.2525, 0.2902, 0.3005, 0.2651, 0.2651, 0.0903, 0.1895, 0.2113, 0.2222, 0.2258, 0.2445, 0.2554, 0.2321, 0.2495, 0.2633, 0.2001, 0.2096, 0.2628, 0.2187, 0.2103, 0.2212, 0.1994, 0.1885, 0.208, 0.2129, 0.2107, 0.2467, 0.2542, 0.2392, 0.1216, 0.1766, 0.1974, 0.2118, 0.2202, 0.2105, 0.1955, 0.1955, 0.2009, 0.2182, 0.2145, 0.2145, 0.1455, 0.2167, 0.2237, 0.2237, 0.2357, 0.2282, 0.2105, 0.2174, 0.2125, 0.1839, 0.1886, 0.1878, 0.1879, 0.1987, 0.2196, 0.2196, 0.1836, 0.2026, 0.1967, 0.2021, 0.2101, 0.2158, 0.2242, 0.219, 0.2213, 0.209, 0.209, 0.202, 0.1678, 0.1621, 0.1996, 0.1896, 0.181, 0.2154, 0.2154, 0.2154, 0.1716, 0.2398, 0.221, 0.1879, 0.1853, 0.1968, 0.2198, 0.1968, 0.1954, 0.1908, 0.1908, 0.1895, 0.1915, 0.1997, 0.1997, 0.2048, 0.2053, 0.1936, 0.2342, 0.2201, 0.1896, 0.1808, 0.2078, 0.2078, 0.2171, 0.2171, 0.1759, 0.1806, 0.189, 0.2136, 0.1988, 0.1946, 0.1995, 0.2112, 0.2131, 0.2145]
loss_objectness: [0.3133, 0.1724, 0.0846, 0.0666, 0.0476, 0.0535, 0.0697, 0.0606, 0.0489, 0.0441, 0.043, 0.0425, 0.0189, 0.0522, 0.0792, 0.16, 0.178, 0.1685, 0.1683, 0.1311, 0.1122, 0.1272, 0.1451, 0.1452, 0.0628, 0.1093, 0.1086, 0.1138, 0.0968, 0.0753, 0.0772, 0.0596, 0.0577, 0.0537, 0.0533, 0.0537, 0.1011, 0.0448, 0.0435, 0.0473, 0.041, 0.041, 0.044, 0.0458, 0.0472, 0.0452, 0.0452, 0.041, 0.0946, 0.0472, 0.0401, 0.0392, 0.037, 0.0317, 0.0316, 0.0412, 0.0382, 0.037, 0.0378, 0.0353, 0.0217, 0.025, 0.0252, 0.0325, 0.0323, 0.031, 0.0306, 0.0248, 0.0236, 0.0242, 0.0218, 0.0218, 0.0186, 0.0203, 0.022, 0.023, 0.0183, 0.0139, 0.0151, 0.0179, 0.015, 0.0155, 0.0155, 0.0147, 0.0261, 0.0147, 0.0147, 0.0172, 0.018, 0.0163, 0.0157, 0.0148, 0.0132, 0.0141, 0.0126, 0.0126, 0.0231, 0.0176, 0.0141, 0.0131, 0.0123, 0.0123, 0.011, 0.011, 0.0159, 0.0168, 0.0126, 0.0129, 0.0069, 0.012, 0.0132, 0.0121, 0.012, 0.0118, 0.0138, 0.0166, 0.0145, 0.0109, 0.0117, 0.014, 0.0118, 0.0131, 0.0114, 0.0101, 0.0125, 0.0125, 0.0125, 0.016, 0.0178, 0.012, 0.0107, 0.0107, 0.0111, 0.0128, 0.0125, 0.0122, 0.0103, 0.0102, 0.0117, 0.0141, 0.0119, 0.0131, 0.0115, 0.0113, 0.0247, 0.0135, 0.012, 0.0107, 0.0133, 0.0113, 0.0129, 0.0132, 0.012, 0.01, 0.0087, 0.0087, 0.0168, 0.0141, 0.0132, 0.0106, 0.0106, 0.0104, 0.0078, 0.009, 0.0094, 0.0094, 0.011, 0.011, 0.0078, 0.0107, 0.0101, 0.0101, 0.0109, 0.0117, 0.0109, 0.0098, 0.0101, 0.0102, 0.0088, 0.0079]
loss_rpn_box_reg: [0.0381, 0.02, 0.0191, 0.0139, 0.014, 0.0218, 0.0244, 0.0175, 0.0122, 0.0196, 0.02, 0.0163, 0.0124, 0.0273, 0.0273, 0.0256, 0.0408, 0.0305, 0.0292, 0.0338, 0.021, 0.0222, 0.0222, 0.0225, 0.0091, 0.0268, 0.028, 0.0254, 0.0232, 0.0267, 0.024, 0.0235, 0.0265, 0.0274, 0.0234, 0.018, 0.045, 0.0273, 0.0234, 0.0217, 0.0217, 0.0182, 0.0172, 0.0172, 0.0216, 0.0232, 0.0188, 0.0193, 0.0195, 0.0195, 0.021, 0.021, 0.0238, 0.0209, 0.0194, 0.0197, 0.0213, 0.0217, 0.0183, 0.0182, 0.0124, 0.0146, 0.0154, 0.0162, 0.014, 0.0175, 0.0192, 0.0164, 0.0197, 0.0217, 0.017, 0.017, 0.0066, 0.0126, 0.0126, 0.0141, 0.015, 0.0146, 0.0173, 0.0166, 0.013, 0.013, 0.0096, 0.0106, 0.0186, 0.0158, 0.0118, 0.0121, 0.0115, 0.011, 0.0108, 0.0103, 0.0103, 0.0127, 0.0147, 0.0147, 0.0095, 0.0103, 0.0125, 0.0138, 0.0119, 0.0106, 0.0138, 0.0138, 0.0133, 0.0131, 0.0105, 0.0105, 0.0064, 0.0103, 0.0096, 0.0085, 0.0138, 0.0138, 0.0119, 0.0149, 0.0109, 0.0097, 0.01, 0.01, 0.0071, 0.0083, 0.0111, 0.0108, 0.0104, 0.0113, 0.0107, 0.0101, 0.0101, 0.0096, 0.0098, 0.0092, 0.0169, 0.0129, 0.0116, 0.0103, 0.0082, 0.0082, 0.0116, 0.0105, 0.0105, 0.0124, 0.0124, 0.0134, 0.009, 0.0115, 0.0115, 0.0107, 0.0101, 0.0101, 0.0118, 0.0099, 0.0099, 0.0101, 0.0082, 0.0085, 0.0128, 0.0102, 0.0102, 0.0103, 0.0099, 0.0092, 0.0096, 0.0122, 0.0101, 0.0087, 0.009, 0.0119, 0.0156, 0.0111, 0.01, 0.009, 0.009, 0.0099, 0.0099, 0.0096, 0.0096, 0.0093, 0.0121, 0.0115]
model_time: []
evaluator_time: []
total_time: []

Model Loading

import pickle

Filename = "FRCNNsgd.pkl"

# Define the file path where you want to save the model
filename = "/content/drive/MyDrive/dataset1/FRCNN1.pth"

# Save the model to the specified file path
torch.save(model.state_dict(), filename)

# Save the Modle to file in the current working directory
with open(Filename, 'wb') as file:
    pickle.dump(model, file)

# Load the Model back from file
with open(Filename, 'rb') as file:
    model = pickle.load(file)
model

FasterRCNN(
  (transform): GeneralizedRCNNTransform(
      Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
      Resize(min_size=(800,), max_size=1333, mode='bilinear')
  )
  (backbone): BackboneWithFPN(
    (body): IntermediateLayerGetter(
      (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)
      (bn1): FrozenBatchNorm2d(64, eps=0.0)
      (relu): ReLU(inplace=True)
      (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)
      (layer1): Sequential(
        (0): Bottleneck(
          (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(64, eps=0.0)
          (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(64, eps=0.0)
          (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(256, eps=0.0)
          (relu): ReLU(inplace=True)
          (downsample): Sequential(
            (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
            (1): FrozenBatchNorm2d(256, eps=0.0)
          )
        )
        (1): Bottleneck(
          (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(64, eps=0.0)
          (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(64, eps=0.0)
          (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(256, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (2): Bottleneck(
          (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(64, eps=0.0)
          (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(64, eps=0.0)
          (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(256, eps=0.0)
          (relu): ReLU(inplace=True)
        )
      )
      (layer2): Sequential(
        (0): Bottleneck(
          (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(128, eps=0.0)
          (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(128, eps=0.0)
          (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(512, eps=0.0)
          (relu): ReLU(inplace=True)
          (downsample): Sequential(
            (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)
            (1): FrozenBatchNorm2d(512, eps=0.0)
          )
        )
        (1): Bottleneck(
          (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(128, eps=0.0)
          (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(128, eps=0.0)
          (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(512, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (2): Bottleneck(
          (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(128, eps=0.0)
          (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(128, eps=0.0)
          (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(512, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (3): Bottleneck(
          (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(128, eps=0.0)
          (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(128, eps=0.0)
          (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(512, eps=0.0)
          (relu): ReLU(inplace=True)
        )
      )
      (layer3): Sequential(
        (0): Bottleneck(
          (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
          (downsample): Sequential(
            (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False)
            (1): FrozenBatchNorm2d(1024, eps=0.0)
          )
        )
        (1): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (2): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (3): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (4): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (5): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
        )
      )
      (layer4): Sequential(
        (0): Bottleneck(
          (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(512, eps=0.0)
          (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(512, eps=0.0)
          (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(2048, eps=0.0)
          (relu): ReLU(inplace=True)
          (downsample): Sequential(
            (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False)
            (1): FrozenBatchNorm2d(2048, eps=0.0)
          )
        )
        (1): Bottleneck(
          (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(512, eps=0.0)
          (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(512, eps=0.0)
          (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(2048, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (2): Bottleneck(
          (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(512, eps=0.0)
          (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(512, eps=0.0)
          (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(2048, eps=0.0)
          (relu): ReLU(inplace=True)
        )
      )
    )
    (fpn): FeaturePyramidNetwork(
      (inner_blocks): ModuleList(
        (0): Conv2dNormActivation(
          (0): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1))
        )
        (1): Conv2dNormActivation(
          (0): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1))
        )
        (2): Conv2dNormActivation(
          (0): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1))
        )
        (3): Conv2dNormActivation(
          (0): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1))
        )
      )
      (layer_blocks): ModuleList(
        (0-3): 4 x Conv2dNormActivation(
          (0): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
        )
      )
      (extra_blocks): LastLevelMaxPool()
    )
  )
  (rpn): RegionProposalNetwork(
    (anchor_generator): AnchorGenerator()
    (head): RPNHead(
      (conv): Sequential(
        (0): Conv2dNormActivation(
          (0): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
          (1): ReLU(inplace=True)
        )
      )
      (cls_logits): Conv2d(256, 3, kernel_size=(1, 1), stride=(1, 1))
      (bbox_pred): Conv2d(256, 12, kernel_size=(1, 1), stride=(1, 1))
    )
  )
  (roi_heads): RoIHeads(
    (box_roi_pool): MultiScaleRoIAlign(featmap_names=['0', '1', '2', '3'], output_size=(7, 7), sampling_ratio=2)
    (box_head): TwoMLPHead(
      (fc6): Linear(in_features=12544, out_features=1024, bias=True)
      (fc7): Linear(in_features=1024, out_features=1024, bias=True)
    )
    (box_predictor): FastRCNNPredictor(
      (cls_score): Linear(in_features=1024, out_features=11, bias=True)
      (bbox_pred): Linear(in_features=1024, out_features=44, bias=True)
    )
  )
)

Validation

import matplotlib.pyplot as plt

# Number of epochs or iterations
epochs = list(range(1, len(metrics["AP"]) + 1))

# Create a figure and axis for plotting
plt.figure(figsize=(10, 6))

# Plotting precision metrics
plt.plot(epochs, metrics["AP"], label='AP [IoU=0.50:0.95]', marker='o')

# Plotting recall metrics
plt.plot(epochs, metrics["AR_1"], label='AR [maxDets=1]', marker='o')
plt.plot(epochs, metrics["AR_10"], label='AR [maxDets=10]', marker='o')
plt.plot(epochs, metrics["AR_100"], label='AR [maxDets=100]', marker='o')

# Adding titles and labels
plt.title('AP and AR Metrics Over Epochs')
plt.xlabel('Epoch')
plt.ylabel('Metric Value')
plt.legend()

# Show the plot
plt.show()

import matplotlib.pyplot as plt

# Number of epochs or iterations
epochs = list(range(1, len(metrics["loss"]) + 1))

# Create a figure and axis for plotting
plt.figure(figsize=(10, 6))

# Plotting all loss metrics
plt.plot(epochs, metrics["loss"], label='Total Loss', marker='o')
plt.plot(epochs, metrics["loss_classifier"], label='Classifier Loss', marker='o')
plt.plot(epochs, metrics["loss_box_reg"], label='Box Reg. Loss', marker='o')
plt.plot(epochs, metrics["loss_objectness"], label='Objectness Loss', marker='o')
plt.plot(epochs, metrics["loss_rpn_box_reg"], label='RPN Box Reg. Loss', marker='o')

# Adding titles and labels
plt.title('Loss Metrics Over Epochs')
plt.xlabel('Epoch')
plt.ylabel('Loss Value')
plt.legend()

# Show the plot
plt.show()

import re

# Define dictionaries to hold your data
metrics = {
    "AR_1": [],
    "AR_10": [],
    "AR_100": [],
    "AR_small": [],
    "AR_medium": [],
    "AR_large": [],
    "AP": [],           # Add AP metric
    "AP_50": [],        # Add AP_50 metric
    "AP_75": [],        # Add AP_75 metric
    "loss": [],         # Add loss metric
    "loss_classifier": [],  # Add loss_classifier metric
    "loss_box_reg": [],     # Add loss_box_reg metric
    "loss_objectness": [],  # Add loss_objectness metric
    "loss_rpn_box_reg": [],  # Add loss_rpn_box_reg metric
    "model_time": [],        # Add model_time metric
    "evaluator_time": [],    # Add evaluator_time metric
    "total_time": []         # Add total_time metric
}

# Regex patterns to extract data
pattern_ar_1 = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area=   all \| maxDets=  1 \] = (\d+\.\d+)")
pattern_ar_10 = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area=   all \| maxDets= 10 \] = (\d+\.\d+)")
pattern_ar_100 = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area=   all \| maxDets=100 \] = (\d+\.\d+)")
pattern_ar_small = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area= small \| maxDets=100 \] = (\d+\.\d+)")
pattern_ar_medium = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area=medium \| maxDets=100 \] = (\d+\.\d+)")
pattern_ar_large = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area= large \| maxDets=100 \] = (\d+\.\d+)")
pattern_ap = re.compile(r"Average Precision\s+\(AP\) @\[ IoU=0\.50:0\.95 \| area=   all \| maxDets=100 \] = (\d+\.\d+)")
pattern_ap_50 = re.compile(r"Average Precision\s+\(AP\) @\[ IoU=0\.50 \| area=   all \| maxDets=100 \] = (\d+\.\d+)")
pattern_ap_75 = re.compile(r"Average Precision\s+\(AP\) @\[ IoU=0\.75 \| area=   all \| maxDets=100 \] = (\d+\.\d+)")
pattern_loss = re.compile(r"loss: (\d+\.\d+)")                  # Pattern for loss
pattern_loss_classifier = re.compile(r"loss_classifier: (\d+\.\d+)")  # Pattern for loss_classifier
pattern_loss_box_reg = re.compile(r"loss_box_reg: (\d+\.\d+)")        # Pattern for loss_box_reg
pattern_loss_objectness = re.compile(r"loss_objectness: (\d+\.\d+)")  # Pattern for loss_objectness
pattern_loss_rpn_box_reg = re.compile(r"loss_rpn_box_reg: (\d+\.\d+)")# Pattern for loss_rpn_box_reg
pattern_model_time = re.compile(r"Model time:\s+(\d+\.\d+)")        # Pattern for model_time
pattern_evaluator_time = re.compile(r"Evaluator time:\s+(\d+\.\d+)")# Pattern for evaluator_time
pattern_total_time = re.compile(r"Total time:\s+(\d+\.\d+)")        # Pattern for total_time

# Read the log file
with open('eva sgd.txt', 'r') as file:
    file_content = file.read()

    # Handling AR matches
    metrics["AR_1"].extend([float(x) for x in pattern_ar_1.findall(file_content)])
    metrics["AR_10"].extend([float(x) for x in pattern_ar_10.findall(file_content)])
    metrics["AR_100"].extend([float(x) for x in pattern_ar_100.findall(file_content)])
    metrics["AR_small"].extend([float(x) for x in pattern_ar_small.findall(file_content)])
    metrics["AR_medium"].extend([float(x) for x in pattern_ar_medium.findall(file_content)])
    metrics["AR_large"].extend([float(x) for x in pattern_ar_large.findall(file_content)])

    # Handling AP matches
    metrics["AP"].extend([float(x) for x in pattern_ap.findall(file_content)])
    metrics["AP_50"].extend([float(x) for x in pattern_ap_50.findall(file_content)])
    metrics["AP_75"].extend([float(x) for x in pattern_ap_75.findall(file_content)])

    # Handling loss matches
    metrics["loss"].extend([float(x) for x in pattern_loss.findall(file_content)])

    # Handling loss_classifier matches
    metrics["loss_classifier"].extend([float(x) for x in pattern_loss_classifier.findall(file_content)])

    # Handling loss_box_reg matches
    metrics["loss_box_reg"].extend([float(x) for x in pattern_loss_box_reg.findall(file_content)])

    # Handling loss_objectness matches
    metrics["loss_objectness"].extend([float(x) for x in pattern_loss_objectness.findall(file_content)])

    # Handling loss_rpn_box_reg matches
    metrics["loss_rpn_box_reg"].extend([float(x) for x in pattern_loss_rpn_box_reg.findall(file_content)])

    # Handling model_time matches
    metrics["model_time"].extend([float(x) for x in pattern_model_time.findall(file_content)])

    # Handling evaluator_time matches
    metrics["evaluator_time"].extend([float(x) for x in pattern_evaluator_time.findall(file_content)])

    # Handling total_time matches
    metrics["total_time"].extend([float(x) for x in pattern_total_time.findall(file_content)])

# Print the collected metrics to verify
for key, value in metrics.items():
    print(f"{key}: {value}")

AR_1: [0.146, 0.173, 0.196, 0.229, 0.228, 0.234, 0.236, 0.234, 0.235, 0.235, 0.236, 0.236, 0.236, 0.236, 0.236]
AR_10: [0.359, 0.444, 0.471, 0.537, 0.54, 0.544, 0.544, 0.543, 0.543, 0.543, 0.544, 0.544, 0.544, 0.544, 0.544]
AR_100: [0.426, 0.511, 0.528, 0.614, 0.615, 0.61, 0.613, 0.613, 0.614, 0.615, 0.615, 0.615, 0.615, 0.615, 0.615]
AR_small: [0.478, 0.442, 0.49, 0.556, 0.561, 0.541, 0.546, 0.549, 0.547, 0.547, 0.548, 0.548, 0.548, 0.548, 0.548]
AR_medium: [0.405, 0.524, 0.513, 0.596, 0.602, 0.597, 0.599, 0.599, 0.601, 0.601, 0.602, 0.602, 0.602, 0.602, 0.602]
AR_large: [0.515, 0.518, 0.582, 0.663, 0.673, 0.678, 0.685, 0.673, 0.673, 0.673, 0.673, 0.673, 0.673, 0.673, 0.673]
AP: [0.298, 0.404, 0.427, 0.534, 0.54, 0.543, 0.548, 0.547, 0.548, 0.548, 0.548, 0.549, 0.549, 0.549, 0.549]
AP_50: []
AP_75: []
loss: [2.501, 2.1539, 1.2824, 1.1755, 0.9232, 0.903, 0.903, 0.8739, 0.7396, 0.7196, 0.6679, 0.6873, 0.9491, 0.6119, 0.5908, 0.5204, 0.4952, 0.4952, 0.4953, 0.5289, 0.512, 0.5059, 0.4005, 0.3574, 0.2529, 0.3099, 0.3128, 0.3433, 0.3896, 0.3336, 0.2878, 0.3079, 0.3821, 0.402, 0.3498, 0.3488, 0.3512, 0.3512, 0.3133, 0.2674, 0.2504, 0.2161, 0.2449, 0.2423, 0.2168, 0.2075, 0.2281, 0.2375, 0.2486, 0.2155, 0.1932, 0.2339, 0.213, 0.221, 0.2511, 0.236, 0.2556, 0.2548, 0.2291, 0.2299, 0.3329, 0.236, 0.2198, 0.2059, 0.2017, 0.2017, 0.2068, 0.2366, 0.2242, 0.2099, 0.231, 0.239, 0.1748, 0.245, 0.2194, 0.2048, 0.1756, 0.1981, 0.2118, 0.2118, 0.2146, 0.2061, 0.2095, 0.2186, 0.3377, 0.2396, 0.1979, 0.1922, 0.216, 0.212, 0.1899, 0.1954, 0.2007, 0.1722, 0.2126, 0.2127, 0.1904, 0.1936, 0.1984, 0.2162, 0.2084, 0.2009, 0.215, 0.215, 0.2228, 0.2226, 0.1706, 0.1645, 0.1976, 0.1904, 0.1882, 0.1947, 0.1974, 0.1927, 0.2107, 0.2091, 0.2091, 0.2247, 0.2146, 0.211, 0.2751, 0.228, 0.1922, 0.2016, 0.1926, 0.1722, 0.199, 0.2036, 0.2014, 0.2014, 0.1887, 0.1882, 0.1635, 0.199, 0.199, 0.2375, 0.2013, 0.167, 0.167, 0.1836, 0.2026, 0.2026, 0.1901, 0.2141, 0.143, 0.2299, 0.2113, 0.177, 0.1848, 0.1868, 0.2124, 0.2013, 0.1894, 0.2155, 0.1985, 0.1981, 0.2508, 0.1942, 0.2218, 0.2237, 0.2167, 0.2058, 0.18, 0.1557, 0.165, 0.1724, 0.2136, 0.2011, 0.1479, 0.2075, 0.1953, 0.1721, 0.2, 0.2065, 0.1814, 0.1878, 0.19, 0.1885, 0.1979, 0.2031]
loss_classifier: [2.0748, 1.6329, 0.7298, 0.6021, 0.4607, 0.4386, 0.4437, 0.4131, 0.3503, 0.3234, 0.2767, 0.2767, 0.4312, 0.203, 0.203, 0.1771, 0.1601, 0.1601, 0.1667, 0.1667, 0.1708, 0.1356, 0.1111, 0.1061, 0.0738, 0.0912, 0.0937, 0.0937, 0.094, 0.076, 0.0715, 0.089, 0.1129, 0.1104, 0.0948, 0.0948, 0.0984, 0.0903, 0.0721, 0.0729, 0.0707, 0.0646, 0.0796, 0.0679, 0.0586, 0.0661, 0.0661, 0.0669, 0.0577, 0.0577, 0.0612, 0.0727, 0.0633, 0.0633, 0.0637, 0.0629, 0.0711, 0.0719, 0.0654, 0.0654, 0.0899, 0.0729, 0.0586, 0.0529, 0.0542, 0.0527, 0.0559, 0.065, 0.0663, 0.0635, 0.0614, 0.0644, 0.0497, 0.0689, 0.0646, 0.0552, 0.0488, 0.0522, 0.0613, 0.0564, 0.0564, 0.0595, 0.0598, 0.0608, 0.0893, 0.0663, 0.0567, 0.0567, 0.0579, 0.0539, 0.0537, 0.0578, 0.057, 0.0577, 0.0583, 0.0603, 0.0511, 0.0489, 0.0493, 0.058, 0.0619, 0.0622, 0.0575, 0.0575, 0.061, 0.0574, 0.0499, 0.0468, 0.0576, 0.0542, 0.0529, 0.0508, 0.0534, 0.0568, 0.0577, 0.0561, 0.0641, 0.0641, 0.0593, 0.0592, 0.0831, 0.0661, 0.0605, 0.059, 0.0549, 0.0491, 0.0547, 0.0572, 0.0511, 0.0546, 0.0589, 0.0546, 0.0396, 0.0505, 0.0505, 0.067, 0.0544, 0.051, 0.051, 0.0537, 0.0572, 0.0541, 0.0549, 0.0549, 0.0349, 0.0665, 0.0605, 0.0544, 0.0544, 0.052, 0.0576, 0.0571, 0.0489, 0.0497, 0.0562, 0.0543, 0.0759, 0.0629, 0.0629, 0.0627, 0.0592, 0.0584, 0.0458, 0.04, 0.045, 0.0468, 0.0553, 0.055, 0.0408, 0.0596, 0.0568, 0.0488, 0.0506, 0.059, 0.0509, 0.0516, 0.0562, 0.0524, 0.0538, 0.0576]
loss_box_reg: [0.2147, 0.2379, 0.2868, 0.3267, 0.3912, 0.3942, 0.4301, 0.4052, 0.3653, 0.3427, 0.3131, 0.34, 0.4331, 0.3582, 0.3284, 0.3169, 0.2924, 0.3116, 0.3224, 0.3381, 0.3021, 0.2882, 0.2583, 0.2464, 0.164, 0.1996, 0.2095, 0.2423, 0.2579, 0.2311, 0.1954, 0.2112, 0.2273, 0.2543, 0.229, 0.2203, 0.2405, 0.2405, 0.2179, 0.1863, 0.1596, 0.1435, 0.1637, 0.1603, 0.1522, 0.1524, 0.1553, 0.1607, 0.1808, 0.141, 0.1371, 0.1525, 0.1423, 0.151, 0.1642, 0.1642, 0.1797, 0.1665, 0.1484, 0.157, 0.2281, 0.1591, 0.1513, 0.1293, 0.1387, 0.1387, 0.1391, 0.1596, 0.1553, 0.141, 0.1553, 0.1553, 0.1187, 0.1692, 0.1513, 0.1435, 0.124, 0.1356, 0.1436, 0.1498, 0.1474, 0.1447, 0.1422, 0.1447, 0.2355, 0.1588, 0.1329, 0.1329, 0.145, 0.1449, 0.1309, 0.1323, 0.1323, 0.1187, 0.1433, 0.1433, 0.1332, 0.1332, 0.1366, 0.1442, 0.1368, 0.127, 0.1517, 0.1531, 0.1561, 0.1549, 0.1162, 0.1145, 0.1314, 0.1302, 0.1255, 0.1255, 0.1315, 0.1315, 0.1355, 0.1366, 0.1418, 0.1488, 0.1465, 0.1438, 0.1825, 0.1609, 0.1324, 0.1326, 0.1342, 0.1142, 0.1351, 0.1417, 0.1399, 0.1412, 0.1298, 0.1339, 0.1188, 0.1402, 0.1385, 0.1542, 0.1448, 0.1161, 0.117, 0.1204, 0.138, 0.1389, 0.1389, 0.1521, 0.1038, 0.157, 0.1419, 0.1148, 0.125, 0.1283, 0.1465, 0.1352, 0.1314, 0.1516, 0.1371, 0.1312, 0.1537, 0.1263, 0.1498, 0.151, 0.144, 0.1418, 0.1195, 0.1138, 0.1137, 0.1282, 0.1544, 0.1402, 0.1007, 0.1381, 0.1373, 0.1191, 0.1414, 0.1524, 0.1188, 0.1317, 0.1317, 0.1268, 0.1322, 0.1392]
loss_objectness: [0.199, 0.199, 0.1409, 0.0952, 0.0787, 0.0389, 0.0233, 0.0335, 0.0278, 0.0169, 0.0202, 0.0206, 0.0605, 0.0207, 0.0156, 0.013, 0.0105, 0.0105, 0.0101, 0.0101, 0.0116, 0.0108, 0.0071, 0.0071, 0.0093, 0.009, 0.0053, 0.0041, 0.005, 0.0036, 0.0022, 0.0021, 0.0038, 0.0042, 0.0056, 0.0059, 0.0026, 0.004, 0.0032, 0.0025, 0.0024, 0.0024, 0.003, 0.003, 0.0022, 0.0014, 0.002, 0.0022, 0.0028, 0.0026, 0.0018, 0.0017, 0.0013, 0.0013, 0.0013, 0.0014, 0.0022, 0.002, 0.0011, 0.0011, 0.0012, 0.002, 0.0018, 0.0017, 0.0009, 0.0011, 0.002, 0.0018, 0.0015, 0.001, 0.0022, 0.0025, 0.0005, 0.0016, 0.0016, 0.0011, 0.0011, 0.0009, 0.0013, 0.0008, 0.0009, 0.0014, 0.0012, 0.0013, 0.0032, 0.0014, 0.001, 0.001, 0.0011, 0.0011, 0.0019, 0.0016, 0.0016, 0.0011, 0.0008, 0.0008, 0.0031, 0.0021, 0.0012, 0.0018, 0.0013, 0.0009, 0.0011, 0.0019, 0.002, 0.0014, 0.001, 0.0007, 0.003, 0.0014, 0.0013, 0.0012, 0.0013, 0.0012, 0.0009, 0.0008, 0.0012, 0.0014, 0.0011, 0.001, 0.0013, 0.0017, 0.0014, 0.0014, 0.0009, 0.0006, 0.0018, 0.0015, 0.0014, 0.0015, 0.0015, 0.0016, 0.0006, 0.0007, 0.001, 0.0018, 0.0014, 0.0013, 0.0012, 0.0008, 0.0008, 0.001, 0.0016, 0.0016, 0.0007, 0.0015, 0.0012, 0.0011, 0.0015, 0.0013, 0.0012, 0.0012, 0.0008, 0.0008, 0.0011, 0.0011, 0.0072, 0.0011, 0.0011, 0.0017, 0.0014, 0.001, 0.0009, 0.0008, 0.0016, 0.0014, 0.0011, 0.0011, 0.0017, 0.0017, 0.0013, 0.0013, 0.0014, 0.0014, 0.0006, 0.0015, 0.0015, 0.0014, 0.0014, 0.0015]
loss_rpn_box_reg: [0.0125, 0.0125, 0.0126, 0.0171, 0.0166, 0.0166, 0.0144, 0.0122, 0.0118, 0.0118, 0.0158, 0.0162, 0.0243, 0.0142, 0.0127, 0.0101, 0.0084, 0.0091, 0.0093, 0.0099, 0.0117, 0.0123, 0.0104, 0.0104, 0.0059, 0.0101, 0.0081, 0.0075, 0.0089, 0.0072, 0.0061, 0.0065, 0.0073, 0.0096, 0.0096, 0.0087, 0.0097, 0.0098, 0.0085, 0.0073, 0.0056, 0.0054, 0.006, 0.0055, 0.0044, 0.0058, 0.0062, 0.0062, 0.0073, 0.003, 0.004, 0.0047, 0.005, 0.0058, 0.0073, 0.0065, 0.0065, 0.0059, 0.0048, 0.0042, 0.0139, 0.0043, 0.0047, 0.0047, 0.0037, 0.0047, 0.0048, 0.0058, 0.0058, 0.0046, 0.005, 0.005, 0.0059, 0.0063, 0.0063, 0.0041, 0.0042, 0.0046, 0.0044, 0.0053, 0.0053, 0.005, 0.005, 0.005, 0.0097, 0.006, 0.0058, 0.0049, 0.0059, 0.0053, 0.0042, 0.0052, 0.0048, 0.0042, 0.0052, 0.005, 0.003, 0.0032, 0.0042, 0.0046, 0.0042, 0.0042, 0.0052, 0.0065, 0.0068, 0.0061, 0.0048, 0.0039, 0.0056, 0.0046, 0.004, 0.004, 0.0043, 0.0052, 0.0054, 0.0043, 0.0053, 0.0055, 0.0054, 0.0053, 0.0083, 0.0055, 0.0047, 0.006, 0.0041, 0.0038, 0.0056, 0.0048, 0.0036, 0.0036, 0.0035, 0.0034, 0.0046, 0.0045, 0.0049, 0.006, 0.0052, 0.0046, 0.0043, 0.004, 0.0053, 0.0053, 0.0058, 0.0065, 0.0036, 0.0043, 0.0047, 0.0047, 0.0047, 0.0041, 0.0057, 0.0057, 0.0042, 0.0044, 0.0056, 0.0056, 0.014, 0.0049, 0.0051, 0.0073, 0.005, 0.005, 0.0039, 0.0031, 0.0032, 0.0035, 0.0049, 0.0049, 0.0047, 0.0062, 0.0049, 0.0049, 0.0059, 0.0051, 0.0037, 0.0042, 0.0052, 0.004, 0.0053, 0.0056]
model_time: []
evaluator_time: []
total_time: []

import matplotlib.pyplot as plt

# Number of epochs or iterations
start_epoch = 1
end_epoch = 20
epochs = list(range(start_epoch, end_epoch + 1))

# Create a figure and axis for plotting
plt.figure(figsize=(10, 6))

# Plotting all loss metrics
plt.plot(epochs, metrics["loss"][:len(epochs)], label='Total Loss', marker='o')
plt.plot(epochs, metrics["loss_classifier"][:len(epochs)], label='Classifier Loss', marker='o')
plt.plot(epochs, metrics["loss_box_reg"][:len(epochs)], label='Box Reg. Loss', marker='o')
plt.plot(epochs, metrics["loss_objectness"][:len(epochs)], label='Objectness Loss', marker='o')
plt.plot(epochs, metrics["loss_rpn_box_reg"][:len(epochs)], label='RPN Box Reg. Loss', marker='o')

# Adding titles and labels
plt.title('Loss Metrics Over Epochs')
plt.xlabel('Epoch')
plt.ylabel('Loss Value')
plt.legend()

# Show the plot
plt.show()

import matplotlib.pyplot as plt

# Number of epochs or iterations
epochs = list(range(1, len(metrics["AR_1"]) + 1))

# Create a figure and axis for plotting
plt.figure(figsize=(10, 6))

# Plotting recall metrics
plt.plot(epochs, metrics["AR_1"], label='AR [maxDets=1]', marker='o')
plt.plot(epochs, metrics["AR_10"], label='AR [maxDets=10]', marker='o')
plt.plot(epochs, metrics["AR_100"], label='AR [maxDets=100]', marker='o')
plt.plot(epochs, metrics["AR_small"], label='AR [area=small]', marker='o')
plt.plot(epochs, metrics["AR_medium"], label='AR [area=medium]', marker='o')
plt.plot(epochs, metrics["AR_large"], label='AR [area=large]', marker='o')

# Adding titles and labels
plt.title('Average Recall (AR) Metrics Over Epochs')
plt.xlabel('Epoch')
plt.ylabel('Recall Value')
plt.legend()

# Show the plot
plt.show()

import matplotlib.pyplot as plt

# Assuming precision_values and recall_values are provided and correctly matched
precision_values = metrics["AP"]  # List of precision values
recall_values = metrics["AR_100"]  # List of recall values corresponding to some IoU or similar metric

# Sort the data by recall since the precision-recall curve expects this.
sorted_indices = sorted(range(len(recall_values)), key=lambda k: recall_values[k])
precision_values = [precision_values[i] for i in sorted_indices]
recall_values = [recall_values[i] for i in sorted_indices]

# To ensure the plot fully spans, check starts and ends
if recall_values[0] > 0:
    recall_values.insert(0, 0)
    precision_values.insert(0, precision_values[0])

if recall_values[-1] < 1:
    recall_values.append(1)
    precision_values.append(precision_values[-1])

# Create the step plot for the precision-recall curve
plt.figure(figsize=(10, 5))
plt.step(recall_values, precision_values, where='post', label='Precision vs. Recall (Purple)', color='red', linewidth=2.5)

plt.xlabel('Recall')
plt.ylabel('Precision')
plt.title('Precision vs. Recall Curve -- Adam and SGD')
plt.xlim(0, 1)
plt.ylim(0, 1)
plt.grid(True)
plt.legend()

# Draw a horizontal line at y=1
plt.axhline(y=1, color='blue', linestyle='-', linewidth=3.5, label='Adam and sgd ')

# Draw a vertical purple line from y=1 to the precision at the first recall step
if recall_values[0] == 0:
    first_non_zero_precision = next(p for p in precision_values if p > 0)
    plt.vlines(x=recall_values[1], ymin=1, ymax=first_non_zero_precision, colors='red', linestyles='-', linewidth=2.5)

# Add a legend to clarify line meanings
plt.legend(title='Legend', bbox_to_anchor=(1.05, 1), loc='upper left')

plt.show()

import re

# Define dictionaries to hold your data
metrics = {
    "AR_1": [],
    "AR_10": [],
    "AR_100": [],
    "AR_small": [],
    "AR_medium": [],
    "AR_large": [],
    "AP": [],           # Add AP metric
    "AP_50": [],        # Add AP_50 metric
    "AP_75": [],        # Add AP_75 metric
    "loss": [],         # Add loss metric
    "loss_classifier": [],  # Add loss_classifier metric
    "loss_box_reg": [],     # Add loss_box_reg metric
    "loss_objectness": [],  # Add loss_objectness metric
    "loss_rpn_box_reg": [],  # Add loss_rpn_box_reg metric
    "model_time": [],        # Add model_time metric
    "evaluator_time": [],    # Add evaluator_time metric
    "total_time": []         # Add total_time metric
}

# Regex patterns to extract data
pattern_ar_1 = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area=   all \| maxDets=  1 \] = (\d+\.\d+)")
pattern_ar_10 = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area=   all \| maxDets= 10 \] = (\d+\.\d+)")
pattern_ar_100 = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area=   all \| maxDets=100 \] = (\d+\.\d+)")
pattern_ar_small = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area= small \| maxDets=100 \] = (\d+\.\d+)")
pattern_ar_medium = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area=medium \| maxDets=100 \] = (\d+\.\d+)")
pattern_ar_large = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area= large \| maxDets=100 \] = (\d+\.\d+)")
pattern_ap = re.compile(r"Average Precision\s+\(AP\) @\[ IoU=0\.50:0\.95 \| area=   all \| maxDets=100 \] = (\d+\.\d+)")
pattern_ap_50 = re.compile(r"Average Precision\s+\(AP\) @\[ IoU=0\.50 \| area=   all \| maxDets=100 \] = (\d+\.\d+)")
pattern_ap_75 = re.compile(r"Average Precision\s+\(AP\) @\[ IoU=0\.75 \| area=   all \| maxDets=100 \] = (\d+\.\d+)")
pattern_loss = re.compile(r"loss: (\d+\.\d+)")                  # Pattern for loss
pattern_loss_classifier = re.compile(r"loss_classifier: (\d+\.\d+)")  # Pattern for loss_classifier
pattern_loss_box_reg = re.compile(r"loss_box_reg: (\d+\.\d+)")        # Pattern for loss_box_reg
pattern_loss_objectness = re.compile(r"loss_objectness: (\d+\.\d+)")  # Pattern for loss_objectness
pattern_loss_rpn_box_reg = re.compile(r"loss_rpn_box_reg: (\d+\.\d+)")# Pattern for loss_rpn_box_reg
pattern_model_time = re.compile(r"Model time:\s+(\d+\.\d+)")        # Pattern for model_time
pattern_evaluator_time = re.compile(r"Evaluator time:\s+(\d+\.\d+)")# Pattern for evaluator_time
pattern_total_time = re.compile(r"Total time:\s+(\d+\.\d+)")        # Pattern for total_time

# Read the log file
with open('adelta.txt', 'r') as file:
    file_content = file.read()

    # Handling AR matches
    metrics["AR_1"].extend([float(x) for x in pattern_ar_1.findall(file_content)])
    metrics["AR_10"].extend([float(x) for x in pattern_ar_10.findall(file_content)])
    metrics["AR_100"].extend([float(x) for x in pattern_ar_100.findall(file_content)])
    metrics["AR_small"].extend([float(x) for x in pattern_ar_small.findall(file_content)])
    metrics["AR_medium"].extend([float(x) for x in pattern_ar_medium.findall(file_content)])
    metrics["AR_large"].extend([float(x) for x in pattern_ar_large.findall(file_content)])

    # Handling AP matches
    metrics["AP"].extend([float(x) for x in pattern_ap.findall(file_content)])
    metrics["AP_50"].extend([float(x) for x in pattern_ap_50.findall(file_content)])
    metrics["AP_75"].extend([float(x) for x in pattern_ap_75.findall(file_content)])

    # Handling loss matches
    metrics["loss"].extend([float(x) for x in pattern_loss.findall(file_content)])

    # Handling loss_classifier matches
    metrics["loss_classifier"].extend([float(x) for x in pattern_loss_classifier.findall(file_content)])

    # Handling loss_box_reg matches
    metrics["loss_box_reg"].extend([float(x) for x in pattern_loss_box_reg.findall(file_content)])

    # Handling loss_objectness matches
    metrics["loss_objectness"].extend([float(x) for x in pattern_loss_objectness.findall(file_content)])

    # Handling loss_rpn_box_reg matches
    metrics["loss_rpn_box_reg"].extend([float(x) for x in pattern_loss_rpn_box_reg.findall(file_content)])

    # Handling model_time matches
    metrics["model_time"].extend([float(x) for x in pattern_model_time.findall(file_content)])

    # Handling evaluator_time matches
    metrics["evaluator_time"].extend([float(x) for x in pattern_evaluator_time.findall(file_content)])

    # Handling total_time matches
    metrics["total_time"].extend([float(x) for x in pattern_total_time.findall(file_content)])

# Print the collected metrics to verify
for key, value in metrics.items():
    print(f"{key}: {value}")

AR_1: [0.001, 0.006, 0.012, 0.013, 0.012, 0.013, 0.013, 0.013, 0.014, 0.014, 0.014, 0.014, 0.014, 0.014, 0.014]
AR_10: [0.011, 0.041, 0.077, 0.082, 0.084, 0.085, 0.085, 0.086, 0.086, 0.086, 0.086, 0.086, 0.086, 0.086, 0.086]
AR_100: [0.016, 0.051, 0.104, 0.114, 0.119, 0.122, 0.122, 0.124, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125]
AR_small: [0.014, 0.107, 0.135, 0.152, 0.14, 0.154, 0.155, 0.156, 0.157, 0.157, 0.15, 0.15, 0.15, 0.15, 0.15]
AR_medium: [0.006, 0.038, 0.139, 0.147, 0.151, 0.154, 0.153, 0.154, 0.155, 0.155, 0.155, 0.155, 0.155, 0.155, 0.155]
AR_large: [0.008, 0.0, 0.025, 0.025, 0.025, 0.025, 0.025, 0.025, 0.025, 0.025, 0.025, 0.025, 0.025, 0.025, 0.025]
AP: [0.001, 0.011, 0.025, 0.026, 0.027, 0.028, 0.028, 0.029, 0.029, 0.029, 0.028, 0.028, 0.028, 0.028, 0.028]
AP_50: []
AP_75: []
loss: [4.0759, 3.36, 3.1378, 2.9602, 2.9602, 2.8197, 2.6611, 2.4501, 2.0663, 1.8681, 1.6345, 1.6345, 1.1351, 1.1702, 1.0701, 0.9767, 0.9272, 0.8314, 0.8384, 0.9086, 0.8851, 0.8026, 0.9113, 0.9113, 0.9163, 0.9163, 0.9086, 0.9709, 0.9518, 0.8419, 0.8895, 0.9628, 0.9628, 0.8832, 0.8006, 0.8006, 1.0471, 1.0512, 1.0315, 0.9557, 0.9115, 0.9662, 0.9121, 0.8367, 1.0318, 0.9981, 0.9981, 1.0118, 1.06, 0.9999, 0.9513, 0.9513, 0.9775, 0.8397, 0.8397, 0.8745, 0.8969, 0.8969, 0.9435, 0.9443, 0.8089, 0.8925, 0.8562, 0.9047, 0.9314, 0.8906, 0.9511, 1.0254, 1.006, 1.0086, 0.9006, 0.9371, 1.2885, 1.0491, 0.9693, 0.9755, 0.9755, 0.9509, 1.0108, 0.8464, 0.8103, 0.9766, 1.032, 1.032, 1.4218, 1.0095, 1.0162, 1.0405, 0.8868, 0.9059, 0.9683, 0.9683, 0.9465, 0.8826, 0.8826, 0.8776, 1.4561, 1.0857, 1.0857, 0.9701, 0.9484, 0.8402, 0.8402, 0.9218, 0.9491, 0.9491, 0.9549, 0.9549, 0.6304, 0.9999, 0.9999, 1.0741, 1.0208, 0.963, 0.9851, 0.9349, 0.882, 0.9438, 0.9698, 0.9841, 0.8985, 0.9647, 0.8716, 0.8716, 0.9804, 0.9462, 0.968, 0.9544, 0.9351, 0.9478, 0.931, 0.9491, 0.755, 0.9419, 1.0021, 0.9841, 0.8633, 0.9035, 0.8914, 0.8592, 0.8975, 0.8975, 0.9923, 0.8953, 1.2637, 1.1292, 0.9303, 0.8486, 0.877, 0.9349, 0.826, 0.8678, 0.9021, 0.9521, 0.9724, 0.9875, 1.0455, 0.9119, 1.1687, 1.0182, 0.8346, 0.837, 0.8682, 0.9294, 0.9831, 0.934, 0.9029, 0.9214, 0.6101, 0.9464, 0.9644, 0.9311, 0.9757, 1.0226, 0.8524, 0.9881, 0.9881, 1.038, 0.9153, 0.9153]
loss_classifier: [2.5687, 2.6055, 2.5849, 2.5095, 2.4297, 2.2737, 2.1325, 1.9114, 1.7363, 1.3998, 1.1006, 1.0552, 0.7006, 0.7006, 0.5925, 0.5344, 0.5014, 0.4376, 0.4454, 0.4537, 0.4499, 0.4275, 0.4669, 0.4463, 0.4803, 0.4786, 0.4786, 0.513, 0.5024, 0.4381, 0.4577, 0.5025, 0.5025, 0.4909, 0.4076, 0.4076, 0.5305, 0.5372, 0.5223, 0.4734, 0.466, 0.5089, 0.4642, 0.4401, 0.5163, 0.4935, 0.5089, 0.5108, 0.5243, 0.5243, 0.4843, 0.4843, 0.4883, 0.4366, 0.4268, 0.4384, 0.483, 0.4673, 0.4673, 0.4772, 0.4197, 0.4646, 0.4325, 0.4699, 0.4938, 0.4345, 0.4656, 0.5239, 0.5202, 0.4859, 0.4366, 0.4366, 0.6427, 0.5128, 0.4798, 0.4994, 0.4994, 0.4764, 0.5083, 0.421, 0.4081, 0.4935, 0.4976, 0.4948, 0.6696, 0.484, 0.484, 0.4935, 0.4536, 0.4536, 0.4741, 0.4857, 0.4834, 0.4435, 0.4553, 0.4504, 0.648, 0.5417, 0.5324, 0.5126, 0.473, 0.4363, 0.4242, 0.4895, 0.4907, 0.4843, 0.4989, 0.4989, 0.3522, 0.5041, 0.4866, 0.5166, 0.5166, 0.4896, 0.5003, 0.4848, 0.4339, 0.4703, 0.4863, 0.4922, 0.4508, 0.4805, 0.456, 0.4485, 0.4722, 0.4719, 0.5065, 0.495, 0.4696, 0.4696, 0.4724, 0.4825, 0.4027, 0.4774, 0.5084, 0.5064, 0.4275, 0.4459, 0.4438, 0.4411, 0.4754, 0.4754, 0.4808, 0.4682, 0.5983, 0.5538, 0.4809, 0.4486, 0.4563, 0.4563, 0.423, 0.4651, 0.4667, 0.4933, 0.4933, 0.5226, 0.5254, 0.4604, 0.5784, 0.4869, 0.4342, 0.4228, 0.4398, 0.4642, 0.4986, 0.4654, 0.4607, 0.4654, 0.2887, 0.4903, 0.4903, 0.4634, 0.4914, 0.5268, 0.4663, 0.4924, 0.5019, 0.5328, 0.4701, 0.4701]
loss_box_reg: [0.2849, 0.2849, 0.1922, 0.158, 0.1765, 0.2803, 0.2608, 0.2562, 0.2166, 0.2123, 0.2507, 0.2609, 0.2811, 0.3579, 0.3021, 0.3021, 0.2982, 0.2701, 0.2669, 0.2953, 0.2936, 0.2936, 0.3101, 0.3214, 0.3495, 0.3385, 0.3239, 0.386, 0.3826, 0.3406, 0.3665, 0.3798, 0.3798, 0.3771, 0.3359, 0.3424, 0.4053, 0.442, 0.4315, 0.3947, 0.3792, 0.4089, 0.3875, 0.369, 0.398, 0.3643, 0.4189, 0.4203, 0.4688, 0.4057, 0.395, 0.374, 0.403, 0.3437, 0.3625, 0.3625, 0.3493, 0.3838, 0.3838, 0.3838, 0.3246, 0.3785, 0.3719, 0.3729, 0.3978, 0.3787, 0.3845, 0.4201, 0.4178, 0.4331, 0.3988, 0.4087, 0.5884, 0.46, 0.4007, 0.3828, 0.4114, 0.4237, 0.437, 0.3655, 0.3511, 0.3786, 0.4053, 0.4036, 0.6648, 0.4434, 0.4568, 0.3845, 0.3749, 0.3783, 0.4048, 0.4071, 0.4071, 0.3466, 0.3637, 0.3742, 0.5642, 0.4551, 0.4345, 0.4303, 0.4157, 0.3572, 0.3395, 0.3836, 0.3861, 0.3844, 0.3844, 0.3844, 0.2296, 0.428, 0.4208, 0.4718, 0.4274, 0.4103, 0.4112, 0.3944, 0.3741, 0.3989, 0.417, 0.4309, 0.3753, 0.4047, 0.3817, 0.3739, 0.4234, 0.3912, 0.41, 0.4145, 0.3985, 0.3947, 0.414, 0.418, 0.3002, 0.406, 0.4273, 0.4191, 0.3595, 0.385, 0.3755, 0.3648, 0.3745, 0.3767, 0.3905, 0.3905, 0.6241, 0.4678, 0.4035, 0.3564, 0.3876, 0.3876, 0.3398, 0.3602, 0.3757, 0.3953, 0.4017, 0.4037, 0.459, 0.381, 0.5096, 0.4253, 0.3361, 0.3397, 0.3525, 0.3764, 0.4131, 0.4099, 0.3724, 0.3794, 0.206, 0.3871, 0.4017, 0.3893, 0.4169, 0.433, 0.3546, 0.3968, 0.3968, 0.4397, 0.3913, 0.39]
loss_objectness: [1.1855, 0.3081, 0.2341, 0.2463, 0.3229, 0.3229, 0.2339, 0.1492, 0.1413, 0.1271, 0.1404, 0.1482, 0.1311, 0.0991, 0.0991, 0.1022, 0.1022, 0.1048, 0.1022, 0.095, 0.0816, 0.0677, 0.0627, 0.0627, 0.0612, 0.0614, 0.0696, 0.0691, 0.0632, 0.0576, 0.0575, 0.0557, 0.0529, 0.0475, 0.0467, 0.0454, 0.0849, 0.0744, 0.0517, 0.0451, 0.0462, 0.0467, 0.0473, 0.0472, 0.0539, 0.0574, 0.0501, 0.0478, 0.0506, 0.0481, 0.0481, 0.0565, 0.0565, 0.0421, 0.0399, 0.0562, 0.0634, 0.0489, 0.0452, 0.044, 0.052, 0.0391, 0.0382, 0.0393, 0.0637, 0.0479, 0.047, 0.0485, 0.0518, 0.0493, 0.0493, 0.0494, 0.0387, 0.0419, 0.0451, 0.0528, 0.0538, 0.0538, 0.0433, 0.0396, 0.035, 0.032, 0.045, 0.0528, 0.066, 0.0545, 0.052, 0.052, 0.0384, 0.0384, 0.0562, 0.0417, 0.0392, 0.0408, 0.0492, 0.0539, 0.2178, 0.0692, 0.0358, 0.0384, 0.0396, 0.0386, 0.0426, 0.0431, 0.0524, 0.0402, 0.0385, 0.0552, 0.0376, 0.0353, 0.0473, 0.0532, 0.0451, 0.0382, 0.0423, 0.0452, 0.0316, 0.0394, 0.0481, 0.0504, 0.0546, 0.0524, 0.0483, 0.0557, 0.0491, 0.0443, 0.0416, 0.0416, 0.0418, 0.0492, 0.0444, 0.0472, 0.043, 0.0463, 0.0445, 0.0389, 0.0427, 0.0444, 0.0444, 0.0479, 0.0491, 0.0581, 0.056, 0.0488, 0.031, 0.0361, 0.0362, 0.0408, 0.0418, 0.0422, 0.0408, 0.041, 0.0441, 0.0518, 0.0518, 0.0518, 0.049, 0.0483, 0.0587, 0.057, 0.0438, 0.0368, 0.0405, 0.0451, 0.0478, 0.05, 0.0532, 0.052, 0.0904, 0.0569, 0.0541, 0.0541, 0.0434, 0.0329, 0.0433, 0.0472, 0.0436, 0.0436, 0.0432, 0.0465]
loss_rpn_box_reg: [0.0368, 0.0216, 0.0178, 0.0178, 0.0201, 0.0216, 0.0205, 0.0181, 0.0146, 0.0156, 0.0193, 0.0211, 0.0222, 0.0222, 0.0203, 0.0179, 0.0179, 0.0159, 0.0159, 0.0155, 0.013, 0.0123, 0.0149, 0.0173, 0.0254, 0.014, 0.0162, 0.0168, 0.0146, 0.0143, 0.0192, 0.0189, 0.0128, 0.013, 0.013, 0.0102, 0.0264, 0.0263, 0.0218, 0.0136, 0.0135, 0.0165, 0.0128, 0.0111, 0.0163, 0.016, 0.0158, 0.0167, 0.0161, 0.0161, 0.016, 0.0195, 0.0212, 0.0129, 0.0093, 0.0174, 0.0193, 0.0137, 0.0137, 0.0137, 0.0127, 0.0127, 0.0129, 0.0143, 0.0147, 0.0133, 0.015, 0.0179, 0.0204, 0.0192, 0.014, 0.017, 0.0187, 0.0158, 0.0154, 0.0178, 0.0154, 0.016, 0.016, 0.0116, 0.0116, 0.0135, 0.0163, 0.0193, 0.0214, 0.015, 0.015, 0.0181, 0.0131, 0.015, 0.0206, 0.0165, 0.0139, 0.0126, 0.0123, 0.0123, 0.0261, 0.0177, 0.0145, 0.0125, 0.0114, 0.0118, 0.0169, 0.0175, 0.0136, 0.0136, 0.016, 0.0171, 0.011, 0.0127, 0.0161, 0.0161, 0.0175, 0.0177, 0.0188, 0.0173, 0.0098, 0.0126, 0.0124, 0.0142, 0.0178, 0.0178, 0.0144, 0.0143, 0.0168, 0.0157, 0.0134, 0.0155, 0.0155, 0.0123, 0.0123, 0.0167, 0.009, 0.0138, 0.0159, 0.0142, 0.0124, 0.0146, 0.0118, 0.0142, 0.0185, 0.018, 0.0173, 0.0153, 0.0103, 0.0126, 0.0159, 0.0159, 0.015, 0.0155, 0.012, 0.0118, 0.0143, 0.0153, 0.0148, 0.0186, 0.0121, 0.0173, 0.0221, 0.0168, 0.011, 0.0137, 0.0125, 0.0128, 0.0153, 0.0152, 0.0149, 0.0129, 0.0251, 0.0206, 0.0206, 0.0179, 0.0167, 0.0153, 0.0134, 0.0161, 0.0161, 0.0171, 0.0161, 0.0154]
model_time: []
evaluator_time: []
total_time: []

# Example: Extracting the average recall for different thresholds
recall_values_Adelta = metrics["AR_100"]  # Let's use AR_100 as an example for IoU thresholds plotting

# Example: Precision vs. Recall (assuming AP data correlates with precision directly at different recalls)
precision_values_Adelta = metrics["AP"]  # Direct extraction for simplicity in this example

import re

# Define dictionaries to hold your data
metrics = {
    "AR_1": [],
    "AR_10": [],
    "AR_100": [],
    "AR_small": [],
    "AR_medium": [],
    "AR_large": [],
    "AP": [],           # Add AP metric
    "AP_50": [],        # Add AP_50 metric
    "AP_75": [],        # Add AP_75 metric
    "loss": [],         # Add loss metric
    "loss_classifier": [],  # Add loss_classifier metric
    "loss_box_reg": [],     # Add loss_box_reg metric
    "loss_objectness": [],  # Add loss_objectness metric
    "loss_rpn_box_reg": [],  # Add loss_rpn_box_reg metric
    "model_time": [],        # Add model_time metric
    "evaluator_time": [],    # Add evaluator_time metric
    "total_time": []         # Add total_time metric
}

# Regex patterns to extract data
pattern_ar_1 = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area=   all \| maxDets=  1 \] = (\d+\.\d+)")
pattern_ar_10 = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area=   all \| maxDets= 10 \] = (\d+\.\d+)")
pattern_ar_100 = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area=   all \| maxDets=100 \] = (\d+\.\d+)")
pattern_ar_small = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area= small \| maxDets=100 \] = (\d+\.\d+)")
pattern_ar_medium = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area=medium \| maxDets=100 \] = (\d+\.\d+)")
pattern_ar_large = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area= large \| maxDets=100 \] = (\d+\.\d+)")
pattern_ap = re.compile(r"Average Precision\s+\(AP\) @\[ IoU=0\.50:0\.95 \| area=   all \| maxDets=100 \] = (\d+\.\d+)")
pattern_ap_50 = re.compile(r"Average Precision\s+\(AP\) @\[ IoU=0\.50 \| area=   all \| maxDets=100 \] = (\d+\.\d+)")
pattern_ap_75 = re.compile(r"Average Precision\s+\(AP\) @\[ IoU=0\.75 \| area=   all \| maxDets=100 \] = (\d+\.\d+)")
pattern_loss = re.compile(r"loss: (\d+\.\d+)")                  # Pattern for loss
pattern_loss_classifier = re.compile(r"loss_classifier: (\d+\.\d+)")  # Pattern for loss_classifier
pattern_loss_box_reg = re.compile(r"loss_box_reg: (\d+\.\d+)")        # Pattern for loss_box_reg
pattern_loss_objectness = re.compile(r"loss_objectness: (\d+\.\d+)")  # Pattern for loss_objectness
pattern_loss_rpn_box_reg = re.compile(r"loss_rpn_box_reg: (\d+\.\d+)")# Pattern for loss_rpn_box_reg
pattern_model_time = re.compile(r"Model time:\s+(\d+\.\d+)")        # Pattern for model_time
pattern_evaluator_time = re.compile(r"Evaluator time:\s+(\d+\.\d+)")# Pattern for evaluator_time
pattern_total_time = re.compile(r"Total time:\s+(\d+\.\d+)")        # Pattern for total_time

# Read the log file
with open('eva adam.txt', 'r') as file:
    file_content = file.read()

    # Handling AR matches
    metrics["AR_1"].extend([float(x) for x in pattern_ar_1.findall(file_content)])
    metrics["AR_10"].extend([float(x) for x in pattern_ar_10.findall(file_content)])
    metrics["AR_100"].extend([float(x) for x in pattern_ar_100.findall(file_content)])
    metrics["AR_small"].extend([float(x) for x in pattern_ar_small.findall(file_content)])
    metrics["AR_medium"].extend([float(x) for x in pattern_ar_medium.findall(file_content)])
    metrics["AR_large"].extend([float(x) for x in pattern_ar_large.findall(file_content)])

    # Handling AP matches
    metrics["AP"].extend([float(x) for x in pattern_ap.findall(file_content)])
    metrics["AP_50"].extend([float(x) for x in pattern_ap_50.findall(file_content)])
    metrics["AP_75"].extend([float(x) for x in pattern_ap_75.findall(file_content)])

    # Handling loss matches
    metrics["loss"].extend([float(x) for x in pattern_loss.findall(file_content)])

    # Handling loss_classifier matches
    metrics["loss_classifier"].extend([float(x) for x in pattern_loss_classifier.findall(file_content)])

    # Handling loss_box_reg matches
    metrics["loss_box_reg"].extend([float(x) for x in pattern_loss_box_reg.findall(file_content)])

    # Handling loss_objectness matches
    metrics["loss_objectness"].extend([float(x) for x in pattern_loss_objectness.findall(file_content)])

    # Handling loss_rpn_box_reg matches
    metrics["loss_rpn_box_reg"].extend([float(x) for x in pattern_loss_rpn_box_reg.findall(file_content)])

    # Handling model_time matches
    metrics["model_time"].extend([float(x) for x in pattern_model_time.findall(file_content)])

    # Handling evaluator_time matches
    metrics["evaluator_time"].extend([float(x) for x in pattern_evaluator_time.findall(file_content)])

    # Handling total_time matches
    metrics["total_time"].extend([float(x) for x in pattern_total_time.findall(file_content)])

# Print the collected metrics to verify
for key, value in metrics.items():
    print(f"{key}: {value}")

AR_1: [0.06, 0.108, 0.111, 0.176, 0.193, 0.192, 0.208, 0.21, 0.21, 0.21, 0.213, 0.215, 0.216, 0.216, 0.216]
AR_10: [0.138, 0.303, 0.311, 0.483, 0.492, 0.501, 0.518, 0.518, 0.522, 0.519, 0.523, 0.524, 0.524, 0.524, 0.524]
AR_100: [0.18, 0.351, 0.356, 0.545, 0.552, 0.558, 0.576, 0.576, 0.576, 0.574, 0.578, 0.579, 0.58, 0.58, 0.58]
AR_small: [0.206, 0.316, 0.289, 0.438, 0.422, 0.448, 0.444, 0.444, 0.445, 0.447, 0.447, 0.448, 0.448, 0.448, 0.448]
AR_medium: [0.199, 0.396, 0.416, 0.588, 0.599, 0.606, 0.623, 0.63, 0.629, 0.622, 0.626, 0.628, 0.629, 0.629, 0.629]
AR_large: [0.088, 0.183, 0.246, 0.441, 0.487, 0.47, 0.492, 0.513, 0.528, 0.518, 0.532, 0.523, 0.523, 0.523, 0.523]
AP: [0.086, 0.209, 0.212, 0.408, 0.429, 0.452, 0.478, 0.482, 0.483, 0.485, 0.488, 0.49, 0.491, 0.491, 0.491]
AP_50: []
AP_75: []
loss: [3.4412, 1.9666, 1.0744, 0.8417, 0.8124, 0.8505, 0.9567, 0.8243, 0.8379, 0.8269, 0.7398, 0.7319, 0.8187, 0.626, 0.6369, 0.6915, 0.7575, 0.7855, 0.7067, 0.6577, 0.602, 0.5815, 0.6848, 0.5815, 0.237, 0.5648, 0.5648, 0.5875, 0.6245, 0.6245, 0.6124, 0.6184, 0.6771, 0.6771, 0.6658, 0.6658, 0.6804, 0.6202, 0.5724, 0.544, 0.4058, 0.4355, 0.4576, 0.4142, 0.3602, 0.3589, 0.3456, 0.3425, 0.5567, 0.3472, 0.3464, 0.3354, 0.3787, 0.3625, 0.358, 0.4037, 0.3903, 0.3678, 0.3634, 0.3103, 0.4193, 0.3269, 0.3544, 0.3695, 0.3503, 0.3338, 0.3008, 0.3433, 0.3123, 0.2546, 0.2826, 0.2938, 0.4073, 0.2835, 0.2868, 0.2868, 0.2755, 0.2693, 0.2315, 0.2315, 0.2799, 0.2819, 0.286, 0.286, 0.4399, 0.2756, 0.2791, 0.2799, 0.24, 0.2466, 0.2689, 0.2628, 0.2451, 0.2572, 0.2582, 0.2582, 0.3998, 0.3032, 0.2519, 0.2395, 0.242, 0.2497, 0.264, 0.2567, 0.2421, 0.2462, 0.2585, 0.2463, 0.1893, 0.2888, 0.2615, 0.2615, 0.2488, 0.2303, 0.2391, 0.2392, 0.2606, 0.2535, 0.25, 0.2444, 0.3089, 0.2657, 0.2605, 0.2682, 0.2551, 0.2229, 0.2212, 0.2255, 0.2585, 0.275, 0.2316, 0.2247, 0.2695, 0.2695, 0.2545, 0.2302, 0.2295, 0.2392, 0.2893, 0.2819, 0.2583, 0.2398, 0.2494, 0.2512, 0.3497, 0.2639, 0.2498, 0.249, 0.249, 0.25, 0.3066, 0.2656, 0.2258, 0.2576, 0.2593, 0.2605, 0.2393, 0.291, 0.2833, 0.2602, 0.2506, 0.2642, 0.2453, 0.2453, 0.2683, 0.2463, 0.2353, 0.2353, 0.341, 0.2914, 0.2498, 0.2601, 0.2729, 0.2713, 0.2197, 0.2562, 0.2723, 0.2576, 0.2431, 0.2379]
loss_classifier: [2.6871, 1.1984, 0.5573, 0.4224, 0.4127, 0.43, 0.43, 0.3951, 0.3951, 0.3935, 0.3464, 0.3661, 0.3262, 0.2835, 0.2835, 0.2926, 0.2951, 0.3421, 0.3367, 0.2653, 0.2453, 0.2143, 0.2656, 0.2348, 0.0976, 0.1963, 0.2199, 0.2595, 0.3, 0.2793, 0.2727, 0.2489, 0.2657, 0.2679, 0.2378, 0.2378, 0.2065, 0.2539, 0.2272, 0.2248, 0.1643, 0.1803, 0.1871, 0.1623, 0.1426, 0.1411, 0.1264, 0.1245, 0.2793, 0.1236, 0.1226, 0.1258, 0.1354, 0.1303, 0.1295, 0.1394, 0.1387, 0.1358, 0.1216, 0.1062, 0.16, 0.139, 0.1277, 0.1277, 0.1212, 0.1079, 0.1079, 0.1082, 0.1012, 0.0837, 0.095, 0.1015, 0.1331, 0.1123, 0.1066, 0.0995, 0.0995, 0.095, 0.086, 0.0908, 0.0908, 0.1034, 0.1034, 0.1024, 0.1619, 0.1017, 0.1017, 0.0968, 0.084, 0.0838, 0.0875, 0.085, 0.0906, 0.0913, 0.095, 0.1008, 0.1506, 0.1042, 0.0926, 0.0878, 0.0923, 0.0996, 0.0942, 0.0847, 0.0903, 0.0924, 0.0851, 0.0792, 0.068, 0.1062, 0.0999, 0.0915, 0.0905, 0.0789, 0.0853, 0.0898, 0.0934, 0.0889, 0.0813, 0.0784, 0.1102, 0.0903, 0.0932, 0.0955, 0.0816, 0.0771, 0.0734, 0.0734, 0.0922, 0.0931, 0.0895, 0.0815, 0.0898, 0.0898, 0.09, 0.0824, 0.0814, 0.0881, 0.0993, 0.0993, 0.0874, 0.0854, 0.0855, 0.0885, 0.1162, 0.0837, 0.0865, 0.0865, 0.0856, 0.0856, 0.0943, 0.0877, 0.0735, 0.0903, 0.0903, 0.0918, 0.0915, 0.1009, 0.1029, 0.0801, 0.0801, 0.0905, 0.0866, 0.0866, 0.0866, 0.0811, 0.0811, 0.085, 0.1312, 0.1028, 0.0872, 0.088, 0.0955, 0.0907, 0.0854, 0.0945, 0.0945, 0.0889, 0.0874, 0.0836]
loss_box_reg: [0.242, 0.3116, 0.3418, 0.351, 0.3445, 0.3445, 0.3505, 0.2874, 0.3641, 0.3835, 0.3031, 0.2775, 0.3955, 0.2487, 0.2845, 0.3091, 0.3427, 0.3934, 0.3516, 0.2943, 0.27, 0.278, 0.3175, 0.2664, 0.1113, 0.2968, 0.3092, 0.3092, 0.3041, 0.3083, 0.2801, 0.2801, 0.3387, 0.3083, 0.331, 0.3307, 0.3854, 0.2812, 0.2722, 0.2523, 0.2174, 0.2227, 0.2408, 0.2172, 0.208, 0.2026, 0.2009, 0.1985, 0.2358, 0.2126, 0.2008, 0.1787, 0.2111, 0.2021, 0.2104, 0.2228, 0.2209, 0.2143, 0.2001, 0.1927, 0.2258, 0.1936, 0.2139, 0.2237, 0.2038, 0.1861, 0.1812, 0.1927, 0.1895, 0.1607, 0.1764, 0.1781, 0.2401, 0.1738, 0.1738, 0.16, 0.1557, 0.1473, 0.1282, 0.1212, 0.181, 0.1826, 0.168, 0.168, 0.2502, 0.1496, 0.1614, 0.1614, 0.1456, 0.1572, 0.1658, 0.1587, 0.1557, 0.154, 0.1645, 0.1679, 0.2245, 0.1803, 0.159, 0.1522, 0.1517, 0.1467, 0.1487, 0.1487, 0.1393, 0.1409, 0.1527, 0.1509, 0.1151, 0.1607, 0.1388, 0.1451, 0.1451, 0.1414, 0.1478, 0.143, 0.1607, 0.1581, 0.1495, 0.1495, 0.1835, 0.1476, 0.1476, 0.1592, 0.1583, 0.1324, 0.1324, 0.1316, 0.1622, 0.1622, 0.1439, 0.1439, 0.1705, 0.1633, 0.1567, 0.1386, 0.1414, 0.148, 0.1679, 0.1679, 0.1529, 0.1406, 0.1472, 0.1508, 0.2069, 0.1679, 0.1379, 0.1386, 0.1379, 0.131, 0.1809, 0.1641, 0.1359, 0.1535, 0.1555, 0.1555, 0.1408, 0.1593, 0.1593, 0.1569, 0.1573, 0.161, 0.1528, 0.1528, 0.1559, 0.1417, 0.1483, 0.1483, 0.1974, 0.1671, 0.1375, 0.1574, 0.1624, 0.1412, 0.1366, 0.1455, 0.165, 0.156, 0.1418, 0.1418]
loss_objectness: [0.4842, 0.1556, 0.1066, 0.0588, 0.0425, 0.0415, 0.0581, 0.0613, 0.0537, 0.0555, 0.0724, 0.0815, 0.0776, 0.0706, 0.0516, 0.0429, 0.0464, 0.0324, 0.0316, 0.0441, 0.0439, 0.0288, 0.0325, 0.0311, 0.0123, 0.0258, 0.0264, 0.0277, 0.0231, 0.0225, 0.0352, 0.0333, 0.0374, 0.0435, 0.0319, 0.0302, 0.0661, 0.0324, 0.0196, 0.0185, 0.0146, 0.0157, 0.0167, 0.0118, 0.0113, 0.0113, 0.0081, 0.0081, 0.0143, 0.0074, 0.0074, 0.0095, 0.0086, 0.0069, 0.0088, 0.0093, 0.0059, 0.0059, 0.0059, 0.0061, 0.0096, 0.0089, 0.0081, 0.0066, 0.0055, 0.006, 0.0064, 0.0056, 0.0042, 0.0051, 0.0046, 0.0046, 0.0098, 0.0055, 0.0049, 0.0046, 0.0041, 0.0033, 0.0033, 0.0036, 0.0036, 0.004, 0.0045, 0.0034, 0.0036, 0.0052, 0.0047, 0.003, 0.003, 0.0029, 0.0025, 0.003, 0.0023, 0.0025, 0.0034, 0.003, 0.0096, 0.0038, 0.0033, 0.0033, 0.0044, 0.0039, 0.0035, 0.0038, 0.0039, 0.0037, 0.0033, 0.0033, 0.0013, 0.0035, 0.004, 0.0042, 0.0039, 0.0034, 0.0037, 0.0033, 0.0022, 0.003, 0.0036, 0.0037, 0.0038, 0.0035, 0.0034, 0.0035, 0.0033, 0.0032, 0.0029, 0.0038, 0.0034, 0.0034, 0.0026, 0.0025, 0.0016, 0.002, 0.0022, 0.0025, 0.0024, 0.0032, 0.0031, 0.0029, 0.0028, 0.0026, 0.0029, 0.0036, 0.006, 0.0025, 0.0034, 0.0047, 0.0039, 0.0031, 0.004, 0.0038, 0.0026, 0.0026, 0.003, 0.003, 0.001, 0.0021, 0.0035, 0.0033, 0.0028, 0.0033, 0.0037, 0.0042, 0.0035, 0.0035, 0.002, 0.0021, 0.0029, 0.0027, 0.0027, 0.003, 0.0032, 0.0025, 0.0021, 0.0028, 0.0033, 0.0022, 0.0022, 0.0022]
loss_rpn_box_reg: [0.0278, 0.0162, 0.0162, 0.0132, 0.0132, 0.0172, 0.0223, 0.0214, 0.0173, 0.0216, 0.0199, 0.0199, 0.0195, 0.0216, 0.0199, 0.0173, 0.0196, 0.0207, 0.0179, 0.0164, 0.0216, 0.0215, 0.0176, 0.0149, 0.0158, 0.0164, 0.0157, 0.0142, 0.0142, 0.0139, 0.0139, 0.018, 0.0227, 0.0217, 0.0217, 0.02, 0.0224, 0.017, 0.0166, 0.0143, 0.0138, 0.0144, 0.0136, 0.0135, 0.0128, 0.0097, 0.0088, 0.0088, 0.0273, 0.0107, 0.009, 0.0081, 0.0104, 0.0091, 0.0093, 0.0093, 0.0095, 0.0107, 0.0103, 0.0099, 0.024, 0.008, 0.0104, 0.0114, 0.0103, 0.0099, 0.0101, 0.0085, 0.0065, 0.007, 0.0077, 0.0076, 0.0243, 0.008, 0.0081, 0.0082, 0.0082, 0.0079, 0.0067, 0.0056, 0.0075, 0.0078, 0.0069, 0.0085, 0.0242, 0.0107, 0.0082, 0.0076, 0.0067, 0.0061, 0.0078, 0.0078, 0.0069, 0.0075, 0.0078, 0.0075, 0.0151, 0.013, 0.008, 0.007, 0.007, 0.007, 0.0069, 0.0069, 0.0065, 0.0062, 0.0065, 0.0062, 0.005, 0.0061, 0.006, 0.0081, 0.0103, 0.0077, 0.0073, 0.0072, 0.0072, 0.0065, 0.006, 0.0062, 0.0114, 0.0075, 0.0076, 0.0078, 0.0069, 0.0054, 0.0052, 0.006, 0.008, 0.008, 0.0062, 0.0065, 0.0075, 0.0095, 0.008, 0.0062, 0.0055, 0.0062, 0.008, 0.0071, 0.0071, 0.0066, 0.007, 0.0072, 0.0205, 0.0052, 0.0062, 0.0075, 0.0075, 0.0074, 0.0091, 0.0085, 0.0066, 0.0069, 0.0068, 0.0065, 0.0061, 0.0082, 0.0085, 0.0079, 0.0074, 0.0074, 0.0067, 0.0079, 0.0085, 0.0064, 0.0069, 0.007, 0.0096, 0.0082, 0.0078, 0.0063, 0.0082, 0.0082, 0.0065, 0.0077, 0.0102, 0.0074, 0.0055, 0.0055]
model_time: []
evaluator_time: []
total_time: []

# Example: Extracting the average recall for different thresholds
recall_values_Adam = metrics["AR_100"]  # Let's use AR_100 as an example for IoU thresholds plotting

# Example: Precision vs. Recall (assuming AP data correlates with precision directly at different recalls)
precision_values_Adam = metrics["AP"]  # Direct extraction for simplicity in this example

import re

# Define dictionaries to hold your data
metrics = {
    "AR_1": [],
    "AR_10": [],
    "AR_100": [],
    "AR_small": [],
    "AR_medium": [],
    "AR_large": [],
    "AP": [],           # Add AP metric
    "AP_50": [],        # Add AP_50 metric
    "AP_75": [],        # Add AP_75 metric
    "loss": [],         # Add loss metric
    "loss_classifier": [],  # Add loss_classifier metric
    "loss_box_reg": [],     # Add loss_box_reg metric
    "loss_objectness": [],  # Add loss_objectness metric
    "loss_rpn_box_reg": [],  # Add loss_rpn_box_reg metric
    "model_time": [],        # Add model_time metric
    "evaluator_time": [],    # Add evaluator_time metric
    "total_time": []         # Add total_time metric
}

# Regex patterns to extract data
pattern_ar_1 = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area=   all \| maxDets=  1 \] = (\d+\.\d+)")
pattern_ar_10 = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area=   all \| maxDets= 10 \] = (\d+\.\d+)")
pattern_ar_100 = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area=   all \| maxDets=100 \] = (\d+\.\d+)")
pattern_ar_small = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area= small \| maxDets=100 \] = (\d+\.\d+)")
pattern_ar_medium = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area=medium \| maxDets=100 \] = (\d+\.\d+)")
pattern_ar_large = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area= large \| maxDets=100 \] = (\d+\.\d+)")
pattern_ap = re.compile(r"Average Precision\s+\(AP\) @\[ IoU=0\.50:0\.95 \| area=   all \| maxDets=100 \] = (\d+\.\d+)")
pattern_ap_50 = re.compile(r"Average Precision\s+\(AP\) @\[ IoU=0\.50 \| area=   all \| maxDets=100 \] = (\d+\.\d+)")
pattern_ap_75 = re.compile(r"Average Precision\s+\(AP\) @\[ IoU=0\.75 \| area=   all \| maxDets=100 \] = (\d+\.\d+)")
pattern_loss = re.compile(r"loss: (\d+\.\d+)")                  # Pattern for loss
pattern_loss_classifier = re.compile(r"loss_classifier: (\d+\.\d+)")  # Pattern for loss_classifier
pattern_loss_box_reg = re.compile(r"loss_box_reg: (\d+\.\d+)")        # Pattern for loss_box_reg
pattern_loss_objectness = re.compile(r"loss_objectness: (\d+\.\d+)")  # Pattern for loss_objectness
pattern_loss_rpn_box_reg = re.compile(r"loss_rpn_box_reg: (\d+\.\d+)")# Pattern for loss_rpn_box_reg
pattern_model_time = re.compile(r"Model time:\s+(\d+\.\d+)")        # Pattern for model_time
pattern_evaluator_time = re.compile(r"Evaluator time:\s+(\d+\.\d+)")# Pattern for evaluator_time
pattern_total_time = re.compile(r"Total time:\s+(\d+\.\d+)")        # Pattern for total_time

# Read the log file
with open('eva adamn_sgd.txt', 'r') as file:
    file_content = file.read()

    # Handling AR matches
    metrics["AR_1"].extend([float(x) for x in pattern_ar_1.findall(file_content)])
    metrics["AR_10"].extend([float(x) for x in pattern_ar_10.findall(file_content)])
    metrics["AR_100"].extend([float(x) for x in pattern_ar_100.findall(file_content)])
    metrics["AR_small"].extend([float(x) for x in pattern_ar_small.findall(file_content)])
    metrics["AR_medium"].extend([float(x) for x in pattern_ar_medium.findall(file_content)])
    metrics["AR_large"].extend([float(x) for x in pattern_ar_large.findall(file_content)])

    # Handling AP matches
    metrics["AP"].extend([float(x) for x in pattern_ap.findall(file_content)])
    metrics["AP_50"].extend([float(x) for x in pattern_ap_50.findall(file_content)])
    metrics["AP_75"].extend([float(x) for x in pattern_ap_75.findall(file_content)])

    # Handling loss matches
    metrics["loss"].extend([float(x) for x in pattern_loss.findall(file_content)])

    # Handling loss_classifier matches
    metrics["loss_classifier"].extend([float(x) for x in pattern_loss_classifier.findall(file_content)])

    # Handling loss_box_reg matches
    metrics["loss_box_reg"].extend([float(x) for x in pattern_loss_box_reg.findall(file_content)])

    # Handling loss_objectness matches
    metrics["loss_objectness"].extend([float(x) for x in pattern_loss_objectness.findall(file_content)])

    # Handling loss_rpn_box_reg matches
    metrics["loss_rpn_box_reg"].extend([float(x) for x in pattern_loss_rpn_box_reg.findall(file_content)])

    # Handling model_time matches
    metrics["model_time"].extend([float(x) for x in pattern_model_time.findall(file_content)])

    # Handling evaluator_time matches
    metrics["evaluator_time"].extend([float(x) for x in pattern_evaluator_time.findall(file_content)])

    # Handling total_time matches
    metrics["total_time"].extend([float(x) for x in pattern_total_time.findall(file_content)])

# Print the collected metrics to verify
for key, value in metrics.items():
    print(f"{key}: {value}")

AR_1: [0.073, 0.007, 0.045, 0.071, 0.105, 0.123, 0.153, 0.157, 0.154, 0.159, 0.153, 0.161, 0.159, 0.166, 0.165]
AR_10: [0.178, 0.056, 0.174, 0.259, 0.315, 0.375, 0.433, 0.442, 0.438, 0.449, 0.447, 0.451, 0.454, 0.455, 0.464]
AR_100: [0.223, 0.101, 0.226, 0.323, 0.371, 0.441, 0.501, 0.506, 0.494, 0.509, 0.509, 0.513, 0.513, 0.517, 0.529]
AR_small: [0.246, 0.115, 0.235, 0.334, 0.35, 0.376, 0.413, 0.434, 0.441, 0.447, 0.455, 0.445, 0.441, 0.45, 0.463]
AR_medium: [0.178, 0.154, 0.278, 0.374, 0.373, 0.504, 0.553, 0.532, 0.533, 0.544, 0.544, 0.55, 0.547, 0.552, 0.569]
AR_large: [0.221, 0.0, 0.09, 0.06, 0.159, 0.237, 0.329, 0.317, 0.307, 0.322, 0.289, 0.339, 0.368, 0.362, 0.375]
AP: [0.098, 0.032, 0.095, 0.173, 0.225, 0.29, 0.347, 0.355, 0.353, 0.361, 0.366, 0.371, 0.372, 0.38, 0.391]
AP_50: []
AP_75: []
loss: [3.3583, 2.3279, 1.209, 0.9238, 0.879, 0.941, 0.8082, 0.6632, 0.5934, 0.8766, 0.7816, 0.6924, 0.5595, 0.9238, 0.9238, 0.8937, 0.8828, 0.8035, 0.7846, 0.7665, 0.7328, 0.8004, 0.8303, 0.8303, 0.5037, 0.7679, 0.7679, 0.6997, 0.6524, 0.6844, 0.6844, 0.7528, 0.8539, 0.8539, 0.6873, 0.6611, 0.9499, 0.8302, 0.6848, 0.7425, 0.7538, 0.6383, 0.6383, 0.5951, 0.6758, 0.6085, 0.5904, 0.6085, 0.9558, 0.7101, 0.6309, 0.6323, 0.6639, 0.575, 0.633, 0.6562, 0.6114, 0.5727, 0.5472, 0.5416, 0.468, 0.4746, 0.5318, 0.6038, 0.581, 0.5266, 0.5092, 0.4859, 0.5286, 0.5525, 0.5126, 0.4816, 0.2004, 0.379, 0.379, 0.3963, 0.4557, 0.4646, 0.4693, 0.4249, 0.4432, 0.4432, 0.3908, 0.3908, 0.5793, 0.3872, 0.3806, 0.4133, 0.3974, 0.35, 0.36, 0.3652, 0.3829, 0.4413, 0.4413, 0.4344, 0.2473, 0.3371, 0.3752, 0.3752, 0.371, 0.371, 0.3706, 0.3706, 0.3642, 0.3939, 0.3779, 0.3779, 0.2582, 0.4245, 0.4, 0.3995, 0.405, 0.3894, 0.3969, 0.4003, 0.3622, 0.3177, 0.341, 0.3467, 0.3358, 0.3452, 0.3529, 0.3529, 0.3148, 0.3396, 0.3396, 0.3545, 0.3728, 0.3848, 0.4097, 0.3852, 0.3552, 0.3552, 0.3827, 0.3661, 0.296, 0.2937, 0.3493, 0.3441, 0.3341, 0.3917, 0.3986, 0.3917, 0.3296, 0.3754, 0.3603, 0.344, 0.334, 0.3332, 0.3676, 0.3456, 0.3438, 0.3364, 0.3178, 0.3118, 0.3092, 0.3466, 0.3585, 0.3632, 0.3632, 0.3369, 0.3831, 0.3831, 0.3283, 0.3066, 0.3514, 0.3514, 0.4111, 0.3929, 0.3124, 0.3136, 0.3399, 0.3399, 0.3414, 0.3355, 0.3644, 0.3639, 0.3578, 0.3658]
loss_classifier: [2.5736, 1.5769, 0.664, 0.4708, 0.4517, 0.4517, 0.3806, 0.3152, 0.3131, 0.3554, 0.3472, 0.3055, 0.2305, 0.3809, 0.4395, 0.4452, 0.4144, 0.3222, 0.3764, 0.3365, 0.3365, 0.3582, 0.3423, 0.3094, 0.2505, 0.263, 0.3391, 0.3233, 0.3024, 0.3101, 0.3198, 0.3344, 0.3837, 0.3838, 0.2968, 0.2968, 0.3417, 0.3417, 0.3111, 0.3458, 0.282, 0.2502, 0.2553, 0.2605, 0.2742, 0.2477, 0.2427, 0.2726, 0.493, 0.2657, 0.2392, 0.2409, 0.275, 0.2283, 0.2227, 0.2589, 0.2353, 0.2379, 0.2376, 0.2112, 0.1478, 0.2071, 0.2175, 0.235, 0.235, 0.2094, 0.1943, 0.1813, 0.1831, 0.1898, 0.1921, 0.1829, 0.085, 0.1573, 0.1573, 0.1639, 0.1723, 0.1774, 0.1746, 0.1655, 0.1613, 0.1654, 0.1517, 0.1499, 0.2718, 0.1535, 0.142, 0.1405, 0.1418, 0.1315, 0.1315, 0.1331, 0.1563, 0.1581, 0.159, 0.152, 0.0932, 0.1366, 0.1365, 0.1365, 0.1305, 0.1323, 0.1415, 0.1418, 0.1418, 0.1465, 0.1397, 0.1397, 0.0994, 0.1346, 0.1377, 0.1344, 0.1295, 0.1551, 0.1575, 0.1554, 0.1343, 0.1343, 0.1387, 0.1387, 0.129, 0.1273, 0.1237, 0.1161, 0.1141, 0.1231, 0.1231, 0.1419, 0.1448, 0.1362, 0.1414, 0.1362, 0.1059, 0.1188, 0.1198, 0.1198, 0.1169, 0.1132, 0.1264, 0.1179, 0.1228, 0.1369, 0.1348, 0.1388, 0.1243, 0.1278, 0.122, 0.122, 0.1333, 0.1333, 0.129, 0.1198, 0.1352, 0.1323, 0.1175, 0.1122, 0.0881, 0.1134, 0.1336, 0.1322, 0.1311, 0.1256, 0.1317, 0.1336, 0.121, 0.1091, 0.1135, 0.1213, 0.1705, 0.1241, 0.1204, 0.1137, 0.113, 0.1125, 0.1187, 0.1128, 0.1185, 0.1278, 0.1257, 0.1278]
loss_box_reg: [0.4333, 0.359, 0.359, 0.3461, 0.3577, 0.3279, 0.3105, 0.2753, 0.2408, 0.3787, 0.3829, 0.3446, 0.2976, 0.42, 0.3396, 0.257, 0.2087, 0.1765, 0.2107, 0.178, 0.207, 0.2739, 0.2363, 0.2845, 0.1813, 0.3321, 0.3321, 0.2844, 0.238, 0.2446, 0.2572, 0.3064, 0.3581, 0.3581, 0.2919, 0.2749, 0.4621, 0.3616, 0.3178, 0.3276, 0.3276, 0.3137, 0.2984, 0.2664, 0.3144, 0.3144, 0.2899, 0.31, 0.3486, 0.3435, 0.293, 0.297, 0.297, 0.2813, 0.2909, 0.3541, 0.2933, 0.2812, 0.2812, 0.2812, 0.2861, 0.2416, 0.2739, 0.289, 0.2642, 0.2642, 0.2546, 0.2525, 0.2902, 0.3005, 0.2651, 0.2651, 0.0903, 0.1895, 0.2113, 0.2222, 0.2258, 0.2445, 0.2554, 0.2321, 0.2495, 0.2633, 0.2001, 0.2096, 0.2628, 0.2187, 0.2103, 0.2212, 0.1994, 0.1885, 0.208, 0.2129, 0.2107, 0.2467, 0.2542, 0.2392, 0.1216, 0.1766, 0.1974, 0.2118, 0.2202, 0.2105, 0.1955, 0.1955, 0.2009, 0.2182, 0.2145, 0.2145, 0.1455, 0.2167, 0.2237, 0.2237, 0.2357, 0.2282, 0.2105, 0.2174, 0.2125, 0.1839, 0.1886, 0.1878, 0.1879, 0.1987, 0.2196, 0.2196, 0.1836, 0.2026, 0.1967, 0.2021, 0.2101, 0.2158, 0.2242, 0.219, 0.2213, 0.209, 0.209, 0.202, 0.1678, 0.1621, 0.1996, 0.1896, 0.181, 0.2154, 0.2154, 0.2154, 0.1716, 0.2398, 0.221, 0.1879, 0.1853, 0.1968, 0.2198, 0.1968, 0.1954, 0.1908, 0.1908, 0.1895, 0.1915, 0.1997, 0.1997, 0.2048, 0.2053, 0.1936, 0.2342, 0.2201, 0.1896, 0.1808, 0.2078, 0.2078, 0.2171, 0.2171, 0.1759, 0.1806, 0.189, 0.2136, 0.1988, 0.1946, 0.1995, 0.2112, 0.2131, 0.2145]
loss_objectness: [0.3133, 0.1724, 0.0846, 0.0666, 0.0476, 0.0535, 0.0697, 0.0606, 0.0489, 0.0441, 0.043, 0.0425, 0.0189, 0.0522, 0.0792, 0.16, 0.178, 0.1685, 0.1683, 0.1311, 0.1122, 0.1272, 0.1451, 0.1452, 0.0628, 0.1093, 0.1086, 0.1138, 0.0968, 0.0753, 0.0772, 0.0596, 0.0577, 0.0537, 0.0533, 0.0537, 0.1011, 0.0448, 0.0435, 0.0473, 0.041, 0.041, 0.044, 0.0458, 0.0472, 0.0452, 0.0452, 0.041, 0.0946, 0.0472, 0.0401, 0.0392, 0.037, 0.0317, 0.0316, 0.0412, 0.0382, 0.037, 0.0378, 0.0353, 0.0217, 0.025, 0.0252, 0.0325, 0.0323, 0.031, 0.0306, 0.0248, 0.0236, 0.0242, 0.0218, 0.0218, 0.0186, 0.0203, 0.022, 0.023, 0.0183, 0.0139, 0.0151, 0.0179, 0.015, 0.0155, 0.0155, 0.0147, 0.0261, 0.0147, 0.0147, 0.0172, 0.018, 0.0163, 0.0157, 0.0148, 0.0132, 0.0141, 0.0126, 0.0126, 0.0231, 0.0176, 0.0141, 0.0131, 0.0123, 0.0123, 0.011, 0.011, 0.0159, 0.0168, 0.0126, 0.0129, 0.0069, 0.012, 0.0132, 0.0121, 0.012, 0.0118, 0.0138, 0.0166, 0.0145, 0.0109, 0.0117, 0.014, 0.0118, 0.0131, 0.0114, 0.0101, 0.0125, 0.0125, 0.0125, 0.016, 0.0178, 0.012, 0.0107, 0.0107, 0.0111, 0.0128, 0.0125, 0.0122, 0.0103, 0.0102, 0.0117, 0.0141, 0.0119, 0.0131, 0.0115, 0.0113, 0.0247, 0.0135, 0.012, 0.0107, 0.0133, 0.0113, 0.0129, 0.0132, 0.012, 0.01, 0.0087, 0.0087, 0.0168, 0.0141, 0.0132, 0.0106, 0.0106, 0.0104, 0.0078, 0.009, 0.0094, 0.0094, 0.011, 0.011, 0.0078, 0.0107, 0.0101, 0.0101, 0.0109, 0.0117, 0.0109, 0.0098, 0.0101, 0.0102, 0.0088, 0.0079]
loss_rpn_box_reg: [0.0381, 0.02, 0.0191, 0.0139, 0.014, 0.0218, 0.0244, 0.0175, 0.0122, 0.0196, 0.02, 0.0163, 0.0124, 0.0273, 0.0273, 0.0256, 0.0408, 0.0305, 0.0292, 0.0338, 0.021, 0.0222, 0.0222, 0.0225, 0.0091, 0.0268, 0.028, 0.0254, 0.0232, 0.0267, 0.024, 0.0235, 0.0265, 0.0274, 0.0234, 0.018, 0.045, 0.0273, 0.0234, 0.0217, 0.0217, 0.0182, 0.0172, 0.0172, 0.0216, 0.0232, 0.0188, 0.0193, 0.0195, 0.0195, 0.021, 0.021, 0.0238, 0.0209, 0.0194, 0.0197, 0.0213, 0.0217, 0.0183, 0.0182, 0.0124, 0.0146, 0.0154, 0.0162, 0.014, 0.0175, 0.0192, 0.0164, 0.0197, 0.0217, 0.017, 0.017, 0.0066, 0.0126, 0.0126, 0.0141, 0.015, 0.0146, 0.0173, 0.0166, 0.013, 0.013, 0.0096, 0.0106, 0.0186, 0.0158, 0.0118, 0.0121, 0.0115, 0.011, 0.0108, 0.0103, 0.0103, 0.0127, 0.0147, 0.0147, 0.0095, 0.0103, 0.0125, 0.0138, 0.0119, 0.0106, 0.0138, 0.0138, 0.0133, 0.0131, 0.0105, 0.0105, 0.0064, 0.0103, 0.0096, 0.0085, 0.0138, 0.0138, 0.0119, 0.0149, 0.0109, 0.0097, 0.01, 0.01, 0.0071, 0.0083, 0.0111, 0.0108, 0.0104, 0.0113, 0.0107, 0.0101, 0.0101, 0.0096, 0.0098, 0.0092, 0.0169, 0.0129, 0.0116, 0.0103, 0.0082, 0.0082, 0.0116, 0.0105, 0.0105, 0.0124, 0.0124, 0.0134, 0.009, 0.0115, 0.0115, 0.0107, 0.0101, 0.0101, 0.0118, 0.0099, 0.0099, 0.0101, 0.0082, 0.0085, 0.0128, 0.0102, 0.0102, 0.0103, 0.0099, 0.0092, 0.0096, 0.0122, 0.0101, 0.0087, 0.009, 0.0119, 0.0156, 0.0111, 0.01, 0.009, 0.009, 0.0099, 0.0099, 0.0096, 0.0096, 0.0093, 0.0121, 0.0115]
model_time: []
evaluator_time: []
total_time: []

# Example: Extracting the average recall for different thresholds
recall_values_ADAMNSGD = metrics["AR_100"]  # Let's use AR_100 as an example for IoU thresholds plotting

# Example: Precision vs. Recall (assuming AP data correlates with precision directly at different recalls)
precision_values_ADAMNSGD = metrics["AP"]  # Direct extraction for simplicity in this example

import re

# Define dictionaries to hold your data
metrics = {
    "AR_1": [],
    "AR_10": [],
    "AR_100": [],
    "AR_small": [],
    "AR_medium": [],
    "AR_large": [],
    "AP": [],           # Add AP metric
    "AP_50": [],        # Add AP_50 metric
    "AP_75": [],        # Add AP_75 metric
    "loss": [],         # Add loss metric
    "loss_classifier": [],  # Add loss_classifier metric
    "loss_box_reg": [],     # Add loss_box_reg metric
    "loss_objectness": [],  # Add loss_objectness metric
    "loss_rpn_box_reg": [],  # Add loss_rpn_box_reg metric
    "model_time": [],        # Add model_time metric
    "evaluator_time": [],    # Add evaluator_time metric
    "total_time": []         # Add total_time metric
}

# Regex patterns to extract data
pattern_ar_1 = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area=   all \| maxDets=  1 \] = (\d+\.\d+)")
pattern_ar_10 = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area=   all \| maxDets= 10 \] = (\d+\.\d+)")
pattern_ar_100 = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area=   all \| maxDets=100 \] = (\d+\.\d+)")
pattern_ar_small = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area= small \| maxDets=100 \] = (\d+\.\d+)")
pattern_ar_medium = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area=medium \| maxDets=100 \] = (\d+\.\d+)")
pattern_ar_large = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area= large \| maxDets=100 \] = (\d+\.\d+)")
pattern_ap = re.compile(r"Average Precision\s+\(AP\) @\[ IoU=0\.50:0\.95 \| area=   all \| maxDets=100 \] = (\d+\.\d+)")
pattern_ap_50 = re.compile(r"Average Precision\s+\(AP\) @\[ IoU=0\.50 \| area=   all \| maxDets=100 \] = (\d+\.\d+)")
pattern_ap_75 = re.compile(r"Average Precision\s+\(AP\) @\[ IoU=0\.75 \| area=   all \| maxDets=100 \] = (\d+\.\d+)")
pattern_loss = re.compile(r"loss: (\d+\.\d+)")                  # Pattern for loss
pattern_loss_classifier = re.compile(r"loss_classifier: (\d+\.\d+)")  # Pattern for loss_classifier
pattern_loss_box_reg = re.compile(r"loss_box_reg: (\d+\.\d+)")        # Pattern for loss_box_reg
pattern_loss_objectness = re.compile(r"loss_objectness: (\d+\.\d+)")  # Pattern for loss_objectness
pattern_loss_rpn_box_reg = re.compile(r"loss_rpn_box_reg: (\d+\.\d+)")# Pattern for loss_rpn_box_reg
pattern_model_time = re.compile(r"Model time:\s+(\d+\.\d+)")        # Pattern for model_time
pattern_evaluator_time = re.compile(r"Evaluator time:\s+(\d+\.\d+)")# Pattern for evaluator_time
pattern_total_time = re.compile(r"Total time:\s+(\d+\.\d+)")        # Pattern for total_time

# Read the log file
with open('eva rsmprob.txt', 'r') as file:
    file_content = file.read()

    # Handling AR matches
    metrics["AR_1"].extend([float(x) for x in pattern_ar_1.findall(file_content)])
    metrics["AR_10"].extend([float(x) for x in pattern_ar_10.findall(file_content)])
    metrics["AR_100"].extend([float(x) for x in pattern_ar_100.findall(file_content)])
    metrics["AR_small"].extend([float(x) for x in pattern_ar_small.findall(file_content)])
    metrics["AR_medium"].extend([float(x) for x in pattern_ar_medium.findall(file_content)])
    metrics["AR_large"].extend([float(x) for x in pattern_ar_large.findall(file_content)])

    # Handling AP matches
    metrics["AP"].extend([float(x) for x in pattern_ap.findall(file_content)])
    metrics["AP_50"].extend([float(x) for x in pattern_ap_50.findall(file_content)])
    metrics["AP_75"].extend([float(x) for x in pattern_ap_75.findall(file_content)])

    # Handling loss matches
    metrics["loss"].extend([float(x) for x in pattern_loss.findall(file_content)])

    # Handling loss_classifier matches
    metrics["loss_classifier"].extend([float(x) for x in pattern_loss_classifier.findall(file_content)])

    # Handling loss_box_reg matches
    metrics["loss_box_reg"].extend([float(x) for x in pattern_loss_box_reg.findall(file_content)])

    # Handling loss_objectness matches
    metrics["loss_objectness"].extend([float(x) for x in pattern_loss_objectness.findall(file_content)])

    # Handling loss_rpn_box_reg matches
    metrics["loss_rpn_box_reg"].extend([float(x) for x in pattern_loss_rpn_box_reg.findall(file_content)])

    # Handling model_time matches
    metrics["model_time"].extend([float(x) for x in pattern_model_time.findall(file_content)])

    # Handling evaluator_time matches
    metrics["evaluator_time"].extend([float(x) for x in pattern_evaluator_time.findall(file_content)])

    # Handling total_time matches
    metrics["total_time"].extend([float(x) for x in pattern_total_time.findall(file_content)])

# Print the collected metrics to verify
for key, value in metrics.items():
    print(f"{key}: {value}")

AR_1: [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
AR_10: [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
AR_100: [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
AR_small: [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
AR_medium: [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
AR_large: [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
AP: [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
AP_50: []
AP_75: []
loss: [3.2039, 1.2988, 1.2517, 1.0549, 1.3226, 2.6148, 2.235, 0.9199, 0.6987, 0.5503, 0.6267, 0.6827, 1.452, 258.9862, 5199.8223, 2652.1675, 1029.4604, 660.6842, 426.3805, 1013.4094, 840.2881, 199.0619, 2028.7905, 2464.8901, 3665.436, 1106.7623, 472.3109, 121.1059, 133.583, 273.1735, 365.2442, 594.5722, 490.288, 517.9962, 480.0167, 315.5242, 304.2431, 272.1476, 265.4117, 169.659, 148.1319, 142.2928, 133.3797, 141.7962, 97.769, 89.6644, 91.671, 91.671, 66.1785, 110.0946, 110.0946, 109.8829, 83.0998, 79.7844, 88.3089, 76.0205, 60.9173, 47.0268, 46.4205, 46.4205, 20.0014, 41.9883, 41.9883, 36.4438, 35.3091, 45.7666, 47.3924, 48.2683, 47.3047, 40.0958, 40.0958, 37.3095, 24.8814, 37.0075, 44.093, 57.8508, 36.9282, 32.768, 47.7588, 51.282, 35.5225, 50.4645, 34.583, 34.583, 87.8664, 47.3212, 39.5393, 38.5752, 42.6587, 50.4157, 34.5905, 29.1215, 35.2129, 37.3594, 38.2908, 48.2291, 15.9957, 32.101, 32.2809, 33.1148, 38.0942, 36.8604, 44.3082, 56.3628, 39.0102, 34.5572, 49.5937, 53.068, 55.8726, 36.021, 36.021, 47.9818, 56.1572, 40.9915, 40.7521, 41.9558, 38.1649, 39.4567, 46.5877, 45.1522, 136.4492, 33.3319, 33.3319, 37.1256, 30.8479, 50.0852, 40.0831, 40.0831, 43.0268, 37.6172, 34.2634, 38.6925, 52.8316, 52.8316, 43.8346, 37.588, 37.4579, 40.121, 42.872, 47.5396, 36.6762, 36.6762, 38.1807, 41.9239, 22.133, 41.8287, 32.3787, 30.8075, 36.132, 38.142, 47.5172, 39.287, 39.6601, 44.6805, 42.819, 42.819, 44.5011, 44.2257, 44.2257, 39.663, 39.663, 37.3944, 37.3944, 48.4363, 42.7073, 35.5755, 38.7166, 38.7166, 7.2644, 40.6111, 41.9498, 44.8585, 54.1296, 45.0768, 36.1728, 41.8349, 35.9192, 33.5233, 41.9988, 41.9988]
loss_classifier: [2.6346, 0.7136, 0.5494, 0.631, 0.7431, 1.1095, 0.6001, 0.3719, 0.2646, 0.2011, 0.2911, 0.3058, 0.6604, 127.6931, 2014.8184, 1474.1609, 611.085, 430.4982, 143.6905, 778.9881, 403.803, 57.0748, 410.3585, 1564.9734, 3275.9192, 562.6231, 244.4862, 41.7808, 69.0693, 69.0693, 124.5114, 283.9956, 126.4839, 175.8158, 99.8471, 89.9003, 84.9447, 39.4093, 43.5565, 35.6502, 14.4577, 12.6049, 10.6868, 8.9238, 9.0629, 10.799, 10.799, 10.799, 2.641, 9.2976, 9.2976, 11.828, 12.7831, 12.8533, 14.3843, 5.5285, 0.5171, 0.2698, 0.2538, 0.2067, 0.1053, 0.2323, 0.292, 0.2598, 0.3117, 0.8353, 0.8437, 0.7768, 0.6391, 0.5869, 0.7374, 0.7374, 0.8751, 0.8751, 0.8575, 0.951, 1.207, 1.207, 1.3659, 1.3659, 1.2666, 1.0201, 1.1253, 1.0971, 0.7757, 1.4586, 1.2743, 0.9598, 0.9598, 1.0065, 0.7998, 0.7629, 0.9716, 0.9789, 1.1156, 1.1156, 0.3973, 1.5712, 0.6473, 0.647, 0.73, 0.9232, 0.9232, 1.0805, 1.2735, 0.9454, 0.6847, 0.7807, 0.9155, 0.7631, 0.5396, 0.5169, 0.6345, 1.1223, 0.7274, 0.5666, 0.9652, 0.6309, 0.5047, 0.5047, 0.7131, 1.4028, 0.758, 0.6061, 0.5191, 0.4484, 0.4169, 0.6742, 0.7473, 0.6455, 0.7205, 1.053, 0.1815, 0.524, 0.5493, 0.9131, 1.0893, 0.604, 0.5619, 0.5431, 0.7054, 0.7671, 0.7514, 0.7198, 0.2661, 0.4315, 0.4703, 0.6102, 0.5346, 0.3968, 0.4551, 0.6406, 0.721, 0.721, 0.4876, 0.7151, 2.7536, 0.5256, 0.7182, 0.7264, 0.5614, 0.7112, 0.8154, 0.8154, 0.7326, 0.7326, 0.6136, 0.5983, 0.4406, 0.4624, 0.5975, 0.9981, 0.9576, 0.4658, 0.5545, 0.5278, 0.5948, 0.6733, 0.7239, 0.6733]
loss_box_reg: [0.2353, 0.2786, 0.269, 0.218, 0.2201, 0.2412, 0.1415, 0.1484, 0.1104, 0.1104, 0.1404, 0.1531, 0.4759, 100.0327, 2918.8621, 1091.9497, 374.0753, 89.7603, 44.3117, 142.968, 67.4619, 25.554, 74.8636, 123.4845, 63.6991, 90.5225, 85.9072, 22.7789, 33.9993, 79.6419, 80.6085, 191.6195, 82.9989, 82.9989, 123.933, 120.3518, 82.5549, 107.1224, 88.7669, 55.9581, 39.1762, 47.2105, 38.1073, 38.1073, 35.096, 29.3037, 29.0876, 25.1067, 28.5581, 35.5208, 35.5208, 24.4844, 21.4768, 15.9272, 14.1213, 4.2127, 0.5805, 0.3479, 0.3883, 0.3031, 0.0933, 0.5077, 0.6207, 0.5993, 0.8865, 0.7469, 0.3043, 0.5655, 0.7198, 0.6141, 0.6148, 0.6622, 1.3191, 1.0455, 0.8651, 0.951, 0.8505, 1.2339, 1.7621, 1.954, 1.891, 0.896, 1.0179, 1.0179, 0.3214, 1.4097, 1.644, 1.0913, 0.6641, 0.9827, 0.9682, 1.5312, 1.5312, 1.5428, 1.6116, 1.1863, 0.4962, 1.5819, 1.5546, 1.1354, 1.1214, 1.1698, 1.137, 1.137, 1.3677, 1.1661, 1.5164, 2.0435, 0.7995, 0.7818, 0.4097, 0.7757, 1.454, 1.543, 2.0573, 1.7006, 1.7006, 0.5337, 0.5314, 0.5314, 1.0747, 2.5008, 1.2173, 0.763, 1.5854, 1.1914, 0.9465, 1.9267, 1.3503, 0.886, 1.0591, 2.4034, 0.4922, 0.9764, 1.1809, 1.4377, 1.4377, 1.2593, 1.2123, 0.9506, 1.4827, 1.1549, 0.8284, 0.8284, 0.2869, 0.7968, 0.9784, 1.3047, 0.4441, 0.6667, 0.6767, 0.7314, 2.0792, 1.5198, 0.6095, 0.6377, 10.0774, 1.6258, 1.5482, 0.8544, 0.7318, 1.7637, 1.9316, 1.5628, 1.3432, 1.3305, 0.5408, 0.5408, 1.2378, 1.9378, 1.6083, 1.9661, 1.7528, 0.6312, 1.5292, 0.67, 0.8445, 0.9395, 0.9588, 0.8385]
loss_objectness: [0.3166, 0.3166, 0.2239, 0.1345, 0.2049, 0.6097, 0.6692, 0.3157, 0.1908, 0.1656, 0.2006, 0.2081, 0.2343, 24.7305, 210.1121, 139.8561, 93.0817, 88.0877, 109.976, 96.8082, 109.258, 109.258, 100.7363, 123.8959, 234.6715, 47.2851, 43.4546, 30.6498, 25.5532, 35.9543, 61.5172, 125.6146, 125.6146, 131.6742, 77.258, 61.8435, 105.421, 75.0196, 75.0196, 66.9722, 67.2271, 55.7569, 46.074, 41.5079, 40.6501, 32.6587, 41.326, 35.2736, 26.6346, 35.9742, 36.0392, 40.7744, 29.0046, 31.7846, 37.6492, 37.6492, 37.019, 36.3957, 34.8001, 34.8001, 17.1295, 31.6889, 31.6889, 26.6947, 24.718, 30.9653, 32.3307, 32.3307, 31.1596, 28.8788, 26.4502, 26.3762, 18.5028, 23.6202, 27.5672, 35.4797, 25.2796, 21.0982, 30.3292, 30.8757, 23.9973, 31.3237, 25.9233, 25.9233, 60.4727, 27.7213, 27.7213, 28.337, 29.4402, 33.9286, 24.795, 19.9744, 23.3236, 25.0347, 23.8722, 25.4885, 11.7689, 18.252, 22.6618, 24.9701, 25.0311, 24.899, 30.955, 33.9098, 26.531, 22.4779, 33.7459, 35.3156, 31.7795, 24.9991, 24.9991, 33.5256, 39.335, 26.8529, 25.3522, 29.7488, 29.121, 29.2802, 24.787, 22.505, 74.9324, 25.7583, 24.6153, 24.8843, 24.2222, 34.0017, 31.6625, 26.0857, 29.3088, 25.9196, 25.9196, 26.831, 30.7382, 33.1904, 29.4245, 24.898, 24.4681, 31.2911, 33.0219, 28.7603, 23.4399, 23.1004, 23.5308, 27.0699, 16.8623, 28.223, 22.4661, 22.5891, 24.0349, 23.1181, 23.1181, 28.402, 28.3379, 28.3379, 29.2609, 29.2609, 26.2172, 27.6522, 29.7883, 26.7828, 26.6561, 24.097, 24.1391, 28.8844, 28.8844, 24.4649, 25.9364, 25.9364, 4.463, 30.6523, 31.3423, 29.2475, 33.1939, 29.7384, 26.3899, 26.3899, 23.4596, 20.6789, 27.4681, 27.847]
loss_rpn_box_reg: [0.0173, 0.0292, 0.0292, 0.0252, 0.0295, 0.0607, 0.1377, 0.0653, 0.0342, 0.0246, 0.0273, 0.0323, 0.0815, 8.8204, 56.0295, 30.526, 24.7837, 20.9473, 24.6669, 24.6669, 32.2513, 20.1335, 39.7727, 52.1484, 91.1466, 10.7598, 14.1286, 12.7459, 10.1752, 12.6669, 25.2533, 37.0607, 35.4058, 31.9363, 23.2427, 14.4586, 31.3225, 12.8808, 13.7246, 25.9141, 28.7024, 22.121, 17.6979, 19.9482, 13.6175, 12.3882, 13.0908, 13.0908, 8.3447, 16.9071, 13.6858, 10.3591, 10.4131, 12.3448, 14.8587, 16.4864, 17.7755, 11.3258, 10.7662, 11.2495, 2.6733, 8.5974, 9.0488, 9.415, 8.4455, 12.1707, 13.1207, 13.373, 13.4515, 12.1153, 9.1422, 9.1422, 4.1844, 12.6925, 13.2039, 15.4171, 10.4051, 5.5364, 9.6681, 14.9676, 8.0572, 11.1851, 8.6166, 7.3149, 26.2966, 14.9454, 8.08, 8.08, 11.0174, 15.4339, 7.9839, 5.5967, 6.5975, 10.9191, 8.4591, 11.4533, 3.3333, 7.2458, 6.8732, 5.726, 8.7146, 8.7146, 9.3121, 16.4104, 7.9361, 8.1952, 12.6412, 15.0755, 22.3781, 5.7209, 5.7209, 10.3965, 10.8813, 7.0965, 6.522, 6.8678, 7.8529, 9.7066, 12.9687, 10.2175, 59.7291, 6.7655, 7.6742, 11.1457, 8.98, 9.6337, 11.9654, 11.173, 10.3651, 8.6694, 8.3527, 8.3527, 21.4197, 15.1997, 9.4829, 8.3418, 8.4908, 8.4908, 13.5257, 14.271, 6.7993, 6.8313, 13.1102, 13.1102, 4.7176, 7.5126, 7.2703, 7.9838, 8.2391, 9.0442, 16.0064, 9.2286, 9.5574, 10.2014, 9.7693, 9.6603, 5.4529, 8.7917, 12.724, 14.2219, 10.6641, 9.6952, 9.6952, 10.0771, 11.1174, 6.0084, 6.3024, 8.0078, 1.123, 9.4268, 9.5961, 12.0252, 14.3342, 12.8331, 8.3081, 8.2428, 8.9517, 9.4732, 9.7129, 11.5158]
model_time: []
evaluator_time: []
total_time: []

# Example: Extracting the average recall for different thresholds
recall_values_RMSprop = metrics["AR_100"]  # Let's use AR_100 as an example for IoU thresholds plotting

# Example: Precision vs. Recall (assuming AP data correlates with precision directly at different recalls)
precision_values_RMSprop = metrics["AP"]  # Direct extraction for simplicity in this example

import re

# Define dictionaries to hold your data
metrics = {
    "AR_1": [],
    "AR_10": [],
    "AR_100": [],
    "AR_small": [],
    "AR_medium": [],
    "AR_large": [],
    "AP": [],           # Add AP metric
    "AP_50": [],        # Add AP_50 metric
    "AP_75": [],        # Add AP_75 metric
    "loss": [],         # Add loss metric
    "loss_classifier": [],  # Add loss_classifier metric
    "loss_box_reg": [],     # Add loss_box_reg metric
    "loss_objectness": [],  # Add loss_objectness metric
    "loss_rpn_box_reg": [],  # Add loss_rpn_box_reg metric
    "model_time": [],        # Add model_time metric
    "evaluator_time": [],    # Add evaluator_time metric
    "total_time": []         # Add total_time metric
}

# Regex patterns to extract data
pattern_ar_1 = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area=   all \| maxDets=  1 \] = (\d+\.\d+)")
pattern_ar_10 = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area=   all \| maxDets= 10 \] = (\d+\.\d+)")
pattern_ar_100 = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area=   all \| maxDets=100 \] = (\d+\.\d+)")
pattern_ar_small = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area= small \| maxDets=100 \] = (\d+\.\d+)")
pattern_ar_medium = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area=medium \| maxDets=100 \] = (\d+\.\d+)")
pattern_ar_large = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area= large \| maxDets=100 \] = (\d+\.\d+)")
pattern_ap = re.compile(r"Average Precision\s+\(AP\) @\[ IoU=0\.50:0\.95 \| area=   all \| maxDets=100 \] = (\d+\.\d+)")
pattern_ap_50 = re.compile(r"Average Precision\s+\(AP\) @\[ IoU=0\.50 \| area=   all \| maxDets=100 \] = (\d+\.\d+)")
pattern_ap_75 = re.compile(r"Average Precision\s+\(AP\) @\[ IoU=0\.75 \| area=   all \| maxDets=100 \] = (\d+\.\d+)")
pattern_loss = re.compile(r"loss: (\d+\.\d+)")                  # Pattern for loss
pattern_loss_classifier = re.compile(r"loss_classifier: (\d+\.\d+)")  # Pattern for loss_classifier
pattern_loss_box_reg = re.compile(r"loss_box_reg: (\d+\.\d+)")        # Pattern for loss_box_reg
pattern_loss_objectness = re.compile(r"loss_objectness: (\d+\.\d+)")  # Pattern for loss_objectness
pattern_loss_rpn_box_reg = re.compile(r"loss_rpn_box_reg: (\d+\.\d+)")# Pattern for loss_rpn_box_reg
pattern_model_time = re.compile(r"Model time:\s+(\d+\.\d+)")        # Pattern for model_time
pattern_evaluator_time = re.compile(r"Evaluator time:\s+(\d+\.\d+)")# Pattern for evaluator_time
pattern_total_time = re.compile(r"Total time:\s+(\d+\.\d+)")        # Pattern for total_time

# Read the log file
with open('eva sgd.txt', 'r') as file:
    file_content = file.read()

    # Handling AR matches
    metrics["AR_1"].extend([float(x) for x in pattern_ar_1.findall(file_content)])
    metrics["AR_10"].extend([float(x) for x in pattern_ar_10.findall(file_content)])
    metrics["AR_100"].extend([float(x) for x in pattern_ar_100.findall(file_content)])
    metrics["AR_small"].extend([float(x) for x in pattern_ar_small.findall(file_content)])
    metrics["AR_medium"].extend([float(x) for x in pattern_ar_medium.findall(file_content)])
    metrics["AR_large"].extend([float(x) for x in pattern_ar_large.findall(file_content)])

    # Handling AP matches
    metrics["AP"].extend([float(x) for x in pattern_ap.findall(file_content)])
    metrics["AP_50"].extend([float(x) for x in pattern_ap_50.findall(file_content)])
    metrics["AP_75"].extend([float(x) for x in pattern_ap_75.findall(file_content)])

    # Handling loss matches
    metrics["loss"].extend([float(x) for x in pattern_loss.findall(file_content)])

    # Handling loss_classifier matches
    metrics["loss_classifier"].extend([float(x) for x in pattern_loss_classifier.findall(file_content)])

    # Handling loss_box_reg matches
    metrics["loss_box_reg"].extend([float(x) for x in pattern_loss_box_reg.findall(file_content)])

    # Handling loss_objectness matches
    metrics["loss_objectness"].extend([float(x) for x in pattern_loss_objectness.findall(file_content)])

    # Handling loss_rpn_box_reg matches
    metrics["loss_rpn_box_reg"].extend([float(x) for x in pattern_loss_rpn_box_reg.findall(file_content)])

    # Handling model_time matches
    metrics["model_time"].extend([float(x) for x in pattern_model_time.findall(file_content)])

    # Handling evaluator_time matches
    metrics["evaluator_time"].extend([float(x) for x in pattern_evaluator_time.findall(file_content)])

    # Handling total_time matches
    metrics["total_time"].extend([float(x) for x in pattern_total_time.findall(file_content)])

# Print the collected metrics to verify
for key, value in metrics.items():
    print(f"{key}: {value}")

AR_1: [0.146, 0.173, 0.196, 0.229, 0.228, 0.234, 0.236, 0.234, 0.235, 0.235, 0.236, 0.236, 0.236, 0.236, 0.236]
AR_10: [0.359, 0.444, 0.471, 0.537, 0.54, 0.544, 0.544, 0.543, 0.543, 0.543, 0.544, 0.544, 0.544, 0.544, 0.544]
AR_100: [0.426, 0.511, 0.528, 0.614, 0.615, 0.61, 0.613, 0.613, 0.614, 0.615, 0.615, 0.615, 0.615, 0.615, 0.615]
AR_small: [0.478, 0.442, 0.49, 0.556, 0.561, 0.541, 0.546, 0.549, 0.547, 0.547, 0.548, 0.548, 0.548, 0.548, 0.548]
AR_medium: [0.405, 0.524, 0.513, 0.596, 0.602, 0.597, 0.599, 0.599, 0.601, 0.601, 0.602, 0.602, 0.602, 0.602, 0.602]
AR_large: [0.515, 0.518, 0.582, 0.663, 0.673, 0.678, 0.685, 0.673, 0.673, 0.673, 0.673, 0.673, 0.673, 0.673, 0.673]
AP: [0.298, 0.404, 0.427, 0.534, 0.54, 0.543, 0.548, 0.547, 0.548, 0.548, 0.548, 0.549, 0.549, 0.549, 0.549]
AP_50: []
AP_75: []
loss: [2.501, 2.1539, 1.2824, 1.1755, 0.9232, 0.903, 0.903, 0.8739, 0.7396, 0.7196, 0.6679, 0.6873, 0.9491, 0.6119, 0.5908, 0.5204, 0.4952, 0.4952, 0.4953, 0.5289, 0.512, 0.5059, 0.4005, 0.3574, 0.2529, 0.3099, 0.3128, 0.3433, 0.3896, 0.3336, 0.2878, 0.3079, 0.3821, 0.402, 0.3498, 0.3488, 0.3512, 0.3512, 0.3133, 0.2674, 0.2504, 0.2161, 0.2449, 0.2423, 0.2168, 0.2075, 0.2281, 0.2375, 0.2486, 0.2155, 0.1932, 0.2339, 0.213, 0.221, 0.2511, 0.236, 0.2556, 0.2548, 0.2291, 0.2299, 0.3329, 0.236, 0.2198, 0.2059, 0.2017, 0.2017, 0.2068, 0.2366, 0.2242, 0.2099, 0.231, 0.239, 0.1748, 0.245, 0.2194, 0.2048, 0.1756, 0.1981, 0.2118, 0.2118, 0.2146, 0.2061, 0.2095, 0.2186, 0.3377, 0.2396, 0.1979, 0.1922, 0.216, 0.212, 0.1899, 0.1954, 0.2007, 0.1722, 0.2126, 0.2127, 0.1904, 0.1936, 0.1984, 0.2162, 0.2084, 0.2009, 0.215, 0.215, 0.2228, 0.2226, 0.1706, 0.1645, 0.1976, 0.1904, 0.1882, 0.1947, 0.1974, 0.1927, 0.2107, 0.2091, 0.2091, 0.2247, 0.2146, 0.211, 0.2751, 0.228, 0.1922, 0.2016, 0.1926, 0.1722, 0.199, 0.2036, 0.2014, 0.2014, 0.1887, 0.1882, 0.1635, 0.199, 0.199, 0.2375, 0.2013, 0.167, 0.167, 0.1836, 0.2026, 0.2026, 0.1901, 0.2141, 0.143, 0.2299, 0.2113, 0.177, 0.1848, 0.1868, 0.2124, 0.2013, 0.1894, 0.2155, 0.1985, 0.1981, 0.2508, 0.1942, 0.2218, 0.2237, 0.2167, 0.2058, 0.18, 0.1557, 0.165, 0.1724, 0.2136, 0.2011, 0.1479, 0.2075, 0.1953, 0.1721, 0.2, 0.2065, 0.1814, 0.1878, 0.19, 0.1885, 0.1979, 0.2031]
loss_classifier: [2.0748, 1.6329, 0.7298, 0.6021, 0.4607, 0.4386, 0.4437, 0.4131, 0.3503, 0.3234, 0.2767, 0.2767, 0.4312, 0.203, 0.203, 0.1771, 0.1601, 0.1601, 0.1667, 0.1667, 0.1708, 0.1356, 0.1111, 0.1061, 0.0738, 0.0912, 0.0937, 0.0937, 0.094, 0.076, 0.0715, 0.089, 0.1129, 0.1104, 0.0948, 0.0948, 0.0984, 0.0903, 0.0721, 0.0729, 0.0707, 0.0646, 0.0796, 0.0679, 0.0586, 0.0661, 0.0661, 0.0669, 0.0577, 0.0577, 0.0612, 0.0727, 0.0633, 0.0633, 0.0637, 0.0629, 0.0711, 0.0719, 0.0654, 0.0654, 0.0899, 0.0729, 0.0586, 0.0529, 0.0542, 0.0527, 0.0559, 0.065, 0.0663, 0.0635, 0.0614, 0.0644, 0.0497, 0.0689, 0.0646, 0.0552, 0.0488, 0.0522, 0.0613, 0.0564, 0.0564, 0.0595, 0.0598, 0.0608, 0.0893, 0.0663, 0.0567, 0.0567, 0.0579, 0.0539, 0.0537, 0.0578, 0.057, 0.0577, 0.0583, 0.0603, 0.0511, 0.0489, 0.0493, 0.058, 0.0619, 0.0622, 0.0575, 0.0575, 0.061, 0.0574, 0.0499, 0.0468, 0.0576, 0.0542, 0.0529, 0.0508, 0.0534, 0.0568, 0.0577, 0.0561, 0.0641, 0.0641, 0.0593, 0.0592, 0.0831, 0.0661, 0.0605, 0.059, 0.0549, 0.0491, 0.0547, 0.0572, 0.0511, 0.0546, 0.0589, 0.0546, 0.0396, 0.0505, 0.0505, 0.067, 0.0544, 0.051, 0.051, 0.0537, 0.0572, 0.0541, 0.0549, 0.0549, 0.0349, 0.0665, 0.0605, 0.0544, 0.0544, 0.052, 0.0576, 0.0571, 0.0489, 0.0497, 0.0562, 0.0543, 0.0759, 0.0629, 0.0629, 0.0627, 0.0592, 0.0584, 0.0458, 0.04, 0.045, 0.0468, 0.0553, 0.055, 0.0408, 0.0596, 0.0568, 0.0488, 0.0506, 0.059, 0.0509, 0.0516, 0.0562, 0.0524, 0.0538, 0.0576]
loss_box_reg: [0.2147, 0.2379, 0.2868, 0.3267, 0.3912, 0.3942, 0.4301, 0.4052, 0.3653, 0.3427, 0.3131, 0.34, 0.4331, 0.3582, 0.3284, 0.3169, 0.2924, 0.3116, 0.3224, 0.3381, 0.3021, 0.2882, 0.2583, 0.2464, 0.164, 0.1996, 0.2095, 0.2423, 0.2579, 0.2311, 0.1954, 0.2112, 0.2273, 0.2543, 0.229, 0.2203, 0.2405, 0.2405, 0.2179, 0.1863, 0.1596, 0.1435, 0.1637, 0.1603, 0.1522, 0.1524, 0.1553, 0.1607, 0.1808, 0.141, 0.1371, 0.1525, 0.1423, 0.151, 0.1642, 0.1642, 0.1797, 0.1665, 0.1484, 0.157, 0.2281, 0.1591, 0.1513, 0.1293, 0.1387, 0.1387, 0.1391, 0.1596, 0.1553, 0.141, 0.1553, 0.1553, 0.1187, 0.1692, 0.1513, 0.1435, 0.124, 0.1356, 0.1436, 0.1498, 0.1474, 0.1447, 0.1422, 0.1447, 0.2355, 0.1588, 0.1329, 0.1329, 0.145, 0.1449, 0.1309, 0.1323, 0.1323, 0.1187, 0.1433, 0.1433, 0.1332, 0.1332, 0.1366, 0.1442, 0.1368, 0.127, 0.1517, 0.1531, 0.1561, 0.1549, 0.1162, 0.1145, 0.1314, 0.1302, 0.1255, 0.1255, 0.1315, 0.1315, 0.1355, 0.1366, 0.1418, 0.1488, 0.1465, 0.1438, 0.1825, 0.1609, 0.1324, 0.1326, 0.1342, 0.1142, 0.1351, 0.1417, 0.1399, 0.1412, 0.1298, 0.1339, 0.1188, 0.1402, 0.1385, 0.1542, 0.1448, 0.1161, 0.117, 0.1204, 0.138, 0.1389, 0.1389, 0.1521, 0.1038, 0.157, 0.1419, 0.1148, 0.125, 0.1283, 0.1465, 0.1352, 0.1314, 0.1516, 0.1371, 0.1312, 0.1537, 0.1263, 0.1498, 0.151, 0.144, 0.1418, 0.1195, 0.1138, 0.1137, 0.1282, 0.1544, 0.1402, 0.1007, 0.1381, 0.1373, 0.1191, 0.1414, 0.1524, 0.1188, 0.1317, 0.1317, 0.1268, 0.1322, 0.1392]
loss_objectness: [0.199, 0.199, 0.1409, 0.0952, 0.0787, 0.0389, 0.0233, 0.0335, 0.0278, 0.0169, 0.0202, 0.0206, 0.0605, 0.0207, 0.0156, 0.013, 0.0105, 0.0105, 0.0101, 0.0101, 0.0116, 0.0108, 0.0071, 0.0071, 0.0093, 0.009, 0.0053, 0.0041, 0.005, 0.0036, 0.0022, 0.0021, 0.0038, 0.0042, 0.0056, 0.0059, 0.0026, 0.004, 0.0032, 0.0025, 0.0024, 0.0024, 0.003, 0.003, 0.0022, 0.0014, 0.002, 0.0022, 0.0028, 0.0026, 0.0018, 0.0017, 0.0013, 0.0013, 0.0013, 0.0014, 0.0022, 0.002, 0.0011, 0.0011, 0.0012, 0.002, 0.0018, 0.0017, 0.0009, 0.0011, 0.002, 0.0018, 0.0015, 0.001, 0.0022, 0.0025, 0.0005, 0.0016, 0.0016, 0.0011, 0.0011, 0.0009, 0.0013, 0.0008, 0.0009, 0.0014, 0.0012, 0.0013, 0.0032, 0.0014, 0.001, 0.001, 0.0011, 0.0011, 0.0019, 0.0016, 0.0016, 0.0011, 0.0008, 0.0008, 0.0031, 0.0021, 0.0012, 0.0018, 0.0013, 0.0009, 0.0011, 0.0019, 0.002, 0.0014, 0.001, 0.0007, 0.003, 0.0014, 0.0013, 0.0012, 0.0013, 0.0012, 0.0009, 0.0008, 0.0012, 0.0014, 0.0011, 0.001, 0.0013, 0.0017, 0.0014, 0.0014, 0.0009, 0.0006, 0.0018, 0.0015, 0.0014, 0.0015, 0.0015, 0.0016, 0.0006, 0.0007, 0.001, 0.0018, 0.0014, 0.0013, 0.0012, 0.0008, 0.0008, 0.001, 0.0016, 0.0016, 0.0007, 0.0015, 0.0012, 0.0011, 0.0015, 0.0013, 0.0012, 0.0012, 0.0008, 0.0008, 0.0011, 0.0011, 0.0072, 0.0011, 0.0011, 0.0017, 0.0014, 0.001, 0.0009, 0.0008, 0.0016, 0.0014, 0.0011, 0.0011, 0.0017, 0.0017, 0.0013, 0.0013, 0.0014, 0.0014, 0.0006, 0.0015, 0.0015, 0.0014, 0.0014, 0.0015]
loss_rpn_box_reg: [0.0125, 0.0125, 0.0126, 0.0171, 0.0166, 0.0166, 0.0144, 0.0122, 0.0118, 0.0118, 0.0158, 0.0162, 0.0243, 0.0142, 0.0127, 0.0101, 0.0084, 0.0091, 0.0093, 0.0099, 0.0117, 0.0123, 0.0104, 0.0104, 0.0059, 0.0101, 0.0081, 0.0075, 0.0089, 0.0072, 0.0061, 0.0065, 0.0073, 0.0096, 0.0096, 0.0087, 0.0097, 0.0098, 0.0085, 0.0073, 0.0056, 0.0054, 0.006, 0.0055, 0.0044, 0.0058, 0.0062, 0.0062, 0.0073, 0.003, 0.004, 0.0047, 0.005, 0.0058, 0.0073, 0.0065, 0.0065, 0.0059, 0.0048, 0.0042, 0.0139, 0.0043, 0.0047, 0.0047, 0.0037, 0.0047, 0.0048, 0.0058, 0.0058, 0.0046, 0.005, 0.005, 0.0059, 0.0063, 0.0063, 0.0041, 0.0042, 0.0046, 0.0044, 0.0053, 0.0053, 0.005, 0.005, 0.005, 0.0097, 0.006, 0.0058, 0.0049, 0.0059, 0.0053, 0.0042, 0.0052, 0.0048, 0.0042, 0.0052, 0.005, 0.003, 0.0032, 0.0042, 0.0046, 0.0042, 0.0042, 0.0052, 0.0065, 0.0068, 0.0061, 0.0048, 0.0039, 0.0056, 0.0046, 0.004, 0.004, 0.0043, 0.0052, 0.0054, 0.0043, 0.0053, 0.0055, 0.0054, 0.0053, 0.0083, 0.0055, 0.0047, 0.006, 0.0041, 0.0038, 0.0056, 0.0048, 0.0036, 0.0036, 0.0035, 0.0034, 0.0046, 0.0045, 0.0049, 0.006, 0.0052, 0.0046, 0.0043, 0.004, 0.0053, 0.0053, 0.0058, 0.0065, 0.0036, 0.0043, 0.0047, 0.0047, 0.0047, 0.0041, 0.0057, 0.0057, 0.0042, 0.0044, 0.0056, 0.0056, 0.014, 0.0049, 0.0051, 0.0073, 0.005, 0.005, 0.0039, 0.0031, 0.0032, 0.0035, 0.0049, 0.0049, 0.0047, 0.0062, 0.0049, 0.0049, 0.0059, 0.0051, 0.0037, 0.0042, 0.0052, 0.004, 0.0053, 0.0056]
model_time: []
evaluator_time: []
total_time: []

# Example: Extracting the average recall for different thresholds
recall_values_SGD = metrics["AR_100"]  # Let's use AR_100 as an example for IoU thresholds plotting

# Example: Precision vs. Recall (assuming AP data correlates with precision directly at different recalls)
precision_values_SGD = metrics["AP"]  # Direct extraction for simplicity in this example

import re

# Define dictionaries to hold your data
metrics = {
    "AR_1": [],
    "AR_10": [],
    "AR_100": [],
    "AR_small": [],
    "AR_medium": [],
    "AR_large": [],
    "AP": [],           # Add AP metric
    "AP_50": [],        # Add AP_50 metric
    "AP_75": [],        # Add AP_75 metric
    "loss": [],         # Add loss metric
    "loss_classifier": [],  # Add loss_classifier metric
    "loss_box_reg": [],     # Add loss_box_reg metric
    "loss_objectness": [],  # Add loss_objectness metric
    "loss_rpn_box_reg": [],  # Add loss_rpn_box_reg metric
    "model_time": [],        # Add model_time metric
    "evaluator_time": [],    # Add evaluator_time metric
    "total_time": []         # Add total_time metric
}

# Regex patterns to extract data
pattern_ar_1 = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area=   all \| maxDets=  1 \] = (\d+\.\d+)")
pattern_ar_10 = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area=   all \| maxDets= 10 \] = (\d+\.\d+)")
pattern_ar_100 = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area=   all \| maxDets=100 \] = (\d+\.\d+)")
pattern_ar_small = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area= small \| maxDets=100 \] = (\d+\.\d+)")
pattern_ar_medium = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area=medium \| maxDets=100 \] = (\d+\.\d+)")
pattern_ar_large = re.compile(r"Average Recall\s+\(AR\) @\[ IoU=0\.50:0\.95 \| area= large \| maxDets=100 \] = (\d+\.\d+)")
pattern_ap = re.compile(r"Average Precision\s+\(AP\) @\[ IoU=0\.50:0\.95 \| area=   all \| maxDets=100 \] = (\d+\.\d+)")
pattern_ap_50 = re.compile(r"Average Precision\s+\(AP\) @\[ IoU=0\.50 \| area=   all \| maxDets=100 \] = (\d+\.\d+)")
pattern_ap_75 = re.compile(r"Average Precision\s+\(AP\) @\[ IoU=0\.75 \| area=   all \| maxDets=100 \] = (\d+\.\d+)")
pattern_loss = re.compile(r"loss: (\d+\.\d+)")                  # Pattern for loss
pattern_loss_classifier = re.compile(r"loss_classifier: (\d+\.\d+)")  # Pattern for loss_classifier
pattern_loss_box_reg = re.compile(r"loss_box_reg: (\d+\.\d+)")        # Pattern for loss_box_reg
pattern_loss_objectness = re.compile(r"loss_objectness: (\d+\.\d+)")  # Pattern for loss_objectness
pattern_loss_rpn_box_reg = re.compile(r"loss_rpn_box_reg: (\d+\.\d+)")# Pattern for loss_rpn_box_reg
pattern_model_time = re.compile(r"Model time:\s+(\d+\.\d+)")        # Pattern for model_time
pattern_evaluator_time = re.compile(r"Evaluator time:\s+(\d+\.\d+)")# Pattern for evaluator_time
pattern_total_time = re.compile(r"Total time:\s+(\d+\.\d+)")        # Pattern for total_time

# Read the log file
with open('eva-sgd-adam.txt', 'r') as file:
    file_content = file.read()

    # Handling AR matches
    metrics["AR_1"].extend([float(x) for x in pattern_ar_1.findall(file_content)])
    metrics["AR_10"].extend([float(x) for x in pattern_ar_10.findall(file_content)])
    metrics["AR_100"].extend([float(x) for x in pattern_ar_100.findall(file_content)])
    metrics["AR_small"].extend([float(x) for x in pattern_ar_small.findall(file_content)])
    metrics["AR_medium"].extend([float(x) for x in pattern_ar_medium.findall(file_content)])
    metrics["AR_large"].extend([float(x) for x in pattern_ar_large.findall(file_content)])

    # Handling AP matches
    metrics["AP"].extend([float(x) for x in pattern_ap.findall(file_content)])
    metrics["AP_50"].extend([float(x) for x in pattern_ap_50.findall(file_content)])
    metrics["AP_75"].extend([float(x) for x in pattern_ap_75.findall(file_content)])

    # Handling loss matches
    metrics["loss"].extend([float(x) for x in pattern_loss.findall(file_content)])

    # Handling loss_classifier matches
    metrics["loss_classifier"].extend([float(x) for x in pattern_loss_classifier.findall(file_content)])

    # Handling loss_box_reg matches
    metrics["loss_box_reg"].extend([float(x) for x in pattern_loss_box_reg.findall(file_content)])

    # Handling loss_objectness matches
    metrics["loss_objectness"].extend([float(x) for x in pattern_loss_objectness.findall(file_content)])

    # Handling loss_rpn_box_reg matches
    metrics["loss_rpn_box_reg"].extend([float(x) for x in pattern_loss_rpn_box_reg.findall(file_content)])

    # Handling model_time matches
    metrics["model_time"].extend([float(x) for x in pattern_model_time.findall(file_content)])

    # Handling evaluator_time matches
    metrics["evaluator_time"].extend([float(x) for x in pattern_evaluator_time.findall(file_content)])

    # Handling total_time matches
    metrics["total_time"].extend([float(x) for x in pattern_total_time.findall(file_content)])

# Print the collected metrics to verify
for key, value in metrics.items():
    print(f"{key}: {value}")

AR_1: [0.155, 0.176, 0.212, 0.201, 0.223, 0.21, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
AR_10: [0.361, 0.44, 0.491, 0.492, 0.536, 0.516, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
AR_100: [0.417, 0.523, 0.561, 0.573, 0.615, 0.59, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
AR_small: [0.397, 0.468, 0.518, 0.549, 0.558, 0.564, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
AR_medium: [0.401, 0.529, 0.536, 0.559, 0.612, 0.575, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
AR_large: [0.512, 0.525, 0.661, 0.591, 0.678, 0.681, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
AP: [0.301, 0.41, 0.462, 0.501, 0.529, 0.515, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
AP_50: []
AP_75: []
loss: [2.4863, 2.3532, 1.1952, 0.9789, 0.9841, 1.1299, 0.9868, 0.8872, 0.7456, 0.6964, 0.6284, 0.6043, 0.4607, 0.4607, 0.5609, 0.4909, 0.47, 0.4464, 0.443, 0.3735, 0.3686, 0.4321, 0.4684, 0.4684, 0.3919, 0.3873, 0.3873, 0.3953, 0.3284, 0.2966, 0.2966, 0.3423, 0.3673, 0.3673, 0.3391, 0.3387, 0.3369, 0.2875, 0.2875, 0.3002, 0.2935, 0.2788, 0.2788, 0.2372, 0.235, 0.2702, 0.2816, 0.3, 0.3058, 0.3005, 0.2973, 0.311, 0.2485, 0.2356, 0.2368, 0.2268, 0.2268, 0.2321, 0.2775, 0.2775, 0.3022, 0.2475, 0.2475, 0.2579, 0.2523, 0.2523, 0.24, 0.2343, 0.2019, 0.2224, 0.2241, 0.2224, 0.2971, 5271547.0, 7212898.5, 1986393.375, 671386.625, 320739.375, 417893.0625, 583216.875, 690144.3125, 311528.8438, 115560.1719, 70980.4688, 6484968.0, 1800933.875, 1776186.625, 617788.625, 2122233.75, 5618370.5, 4074645.75, 51039480.0, 112068888.0, 28259246.0, 28827070.0, 36390144.0, 37082226688.0, 143327376.0, 34987936.0, 22143708.0, 16452572.0, 3729538.75, 3180220.25, 116248224.0, 116248224.0, 47838352.0, 59788872.0, 52835160.0, 52720724.0, 195219552.0, 399970272.0, 395688192.0, 241189632.0, 568194368.0, 271617408.0, 171460368.0, 4976686080.0, 7802124288.0, 61865938944.0, 46775721984.0, 1418080384.0, 31926710272.0, 41220120576.0, 41220120576.0, 102345703424.0, 87810138112.0, 31108685824.0, 43317927936.0, 56634617856.0, 16804518912.0, 10109952000.0, 13764925440.0, 6116950016.0, 205510377472.0, 68524650496.0, 10105732096.0, 18252888064.0, 26050062336.0, 31898882048.0, 245992423424.0, 223513378816.0, 26827456512.0, 80096428032.0, 137168797696.0, 14640758128640.0, 589142556672.0, 205823967232.0, 15113707520.0, 6404826624.0, 6520471552.0, 4512242176.0, 12624415744.0, 29912176640.0, 55517057024.0, 59984121856.0, 60857618432.0, 25029279744.0, 14850866176.0, 12864126976.0, 7330747904.0, 3582105600.0, 2027425024.0, 1145666688.0, 559781504.0, 436397184.0, 368620544.0, 629898304.0, 669884352.0, 543165824.0, 543165824.0, 462647616.0, 253432448.0, 225819984.0, 183020400.0, 185756608.0, 234713504.0, 445998784.0, 5967999488.0, 8547352576.0, 8309713920.0]
loss_classifier: [2.2196, 1.7739, 0.7093, 0.4751, 0.472, 0.5273, 0.4433, 0.3999, 0.3241, 0.2928, 0.2484, 0.2494, 0.2299, 0.1716, 0.2009, 0.1881, 0.1476, 0.1311, 0.1287, 0.1132, 0.1228, 0.1619, 0.1619, 0.1619, 0.1312, 0.1226, 0.1133, 0.1046, 0.0813, 0.0813, 0.0899, 0.0982, 0.0986, 0.0901, 0.0993, 0.1035, 0.0912, 0.0764, 0.0786, 0.0796, 0.0807, 0.0864, 0.0765, 0.0662, 0.0586, 0.072, 0.0751, 0.0751, 0.0823, 0.0708, 0.0708, 0.0752, 0.0614, 0.059, 0.0565, 0.0545, 0.0595, 0.0647, 0.0723, 0.0692, 0.0799, 0.0603, 0.0626, 0.0638, 0.0559, 0.0597, 0.0597, 0.0588, 0.0556, 0.0576, 0.0566, 0.0558, 0.087, 324243.6875, 1578566.875, 640492.75, 347828.1875, 67948.3516, 90915.1875, 123241.5078, 176769.7812, 54361.0312, 29951.9961, 15329.9561, 2.3239, 2.4392, 2.4392, 19.9234, 24.8629, 2.4269, 2.4045, 5607205.0, 4182530.25, 250618.8594, 2.4788, 16860.0645, 2.3647, 1758.8846, 2.361, 2.3722, 2.3722, 2.3534, 2.3452, 2.3225, 2.3834, 2.4005, 2.3685, 2.3961, 11789422.0, 2.4235, 2.3975, 2.342, 2.2743, 2.2749, 2.3054, 2.3037, 2.3533, 2.4123, 2.3364, 2.3036, 101655960.0, 3605993.0, 2.4454, 2.313, 2.2816, 2.3102, 2.2372, 2.2508, 2.306, 2.306, 2.2838, 2.3195, 2.3194, 2.3194, 2.2675, 2.2732, 2.3519, 2.22, 2.2632, 2.3389, 2.3131, 2.2631, 2.2955, 2.2692, 523177787392.0, 202637456.0, 2.2525, 2.3488, 2.3488, 2.2346, 2.1972, 2.247, 2.2059, 2.1549, 2.2604, 2.3344, 2.2541, 2.2808, 2.2808, 2.2059, 2.182, 2.2192, 2.2192, 2.1941, 2.1895, 2.2018, 2.2018, 2.1526, 2.4773, 2.1827, 2.2695, 2.3716, 2.3258, 2.325, 2.1143, 2.1625, 2.2757, 2.2456, 2.1364, 2.1364]
loss_box_reg: [0.181, 0.3038, 0.3343, 0.3329, 0.3406, 0.4682, 0.4586, 0.4185, 0.3841, 0.3433, 0.3362, 0.3362, 0.2125, 0.254, 0.3167, 0.3065, 0.3019, 0.2988, 0.2808, 0.2591, 0.2189, 0.2561, 0.2748, 0.2769, 0.2455, 0.2455, 0.2537, 0.2453, 0.2054, 0.196, 0.196, 0.2213, 0.2694, 0.2589, 0.2389, 0.2371, 0.2336, 0.1991, 0.1962, 0.2012, 0.2132, 0.195, 0.195, 0.1719, 0.1692, 0.1832, 0.2014, 0.2118, 0.2145, 0.2186, 0.2186, 0.2214, 0.1798, 0.1688, 0.1688, 0.1488, 0.1541, 0.1663, 0.1817, 0.1817, 0.2021, 0.1758, 0.1775, 0.1809, 0.1806, 0.1693, 0.1699, 0.1775, 0.1487, 0.1594, 0.1616, 0.1594, 0.2007, 5142962.5, 5142962.5, 1267061.75, 386138.875, 199476.4375, 320711.4688, 434341.1562, 484874.0, 196912.8281, 55739.7344, 29164.416, 0.0179, 0.7598, 0.7598, 0.1743, 31.9085, 163513.4219, 0.0399, 13121892.0, 5030653.0, 213943.625, 739.4225, 64643.7773, 0.0079, 3567.8777, 0.0114, 0.0062, 0.0094, 0.0036, 0.0021, 0.002, 0.0034, 0.0048, 0.0063, 0.0032, 1178148.5, 0.0045, 0.0034, 0.0025, 0.0016, 0.0012, 0.0018, 0.0018, 0.0013, 0.001, 0.0012, 0.0013, 2866243.75, 558154.4375, 0.0069, 0.003, 0.002, 0.0013, 0.0008, 0.001, 0.001, 0.0016, 0.002, 0.002, 0.001, 0.0011, 0.001, 0.0002, 0.0001, 0.0003, 0.001, 0.0013, 0.0016, 0.0033, 0.0042, 0.0042, 154370490368.0, 0.0045, 0.0013, 0.0013, 0.0012, 0.0007, 0.0004, 0.0006, 0.0007, 0.0007, 0.0015, 0.0015, 0.0005, 0.0005, 0.0006, 0.0017, 0.0025, 0.0015, 0.0025, 0.0028, 0.002, 0.0005, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0, 0.0001, 0.0002, 0.0002, 0.0, 0.0, 0.0008, 0.002, 0.002]
loss_objectness: [0.0736, 0.2685, 0.109, 0.0847, 0.0753, 0.0493, 0.0371, 0.0316, 0.0186, 0.0179, 0.0201, 0.02, 0.014, 0.0145, 0.0179, 0.0124, 0.0086, 0.0099, 0.0087, 0.0067, 0.0098, 0.0115, 0.0112, 0.0117, 0.0062, 0.007, 0.0067, 0.0055, 0.0054, 0.0039, 0.0047, 0.005, 0.0055, 0.0033, 0.0039, 0.005, 0.0027, 0.0026, 0.003, 0.0022, 0.0019, 0.0023, 0.0027, 0.0017, 0.0014, 0.0014, 0.0026, 0.0033, 0.004, 0.0016, 0.0017, 0.0015, 0.0012, 0.0008, 0.0008, 0.0012, 0.0017, 0.0015, 0.0015, 0.0015, 0.0024, 0.0023, 0.0023, 0.0017, 0.0015, 0.0012, 0.0014, 0.001, 0.0006, 0.0011, 0.0013, 0.0014, 0.0012, 40562.7852, 52229.1914, 48171.1367, 20998.1211, 6489.3223, 9490.4365, 8365.3896, 7892.6016, 6596.0479, 4572.4634, 5286.834, 4961285.5, 677858.625, 555855.0, 292421.375, 509012.5, 3614144.0, 1587095.875, 3438756.0, 10105214.0, 8628231.0, 8628231.0, 12218656.0, 20404789248.0, 76122056.0, 17763510.0, 11812624.0, 3748131.75, 1942666.0, 824644.1875, 64077120.0, 64077120.0, 11815075.0, 12146524.0, 6192073.0, 13265775.0, 27572750.0, 54201908.0, 72947248.0, 72947248.0, 100719768.0, 152921584.0, 152921584.0, 3005594624.0, 3005594624.0, 15097151488.0, 15097151488.0, 370253568.0, 12318129152.0, 12318129152.0, 15383909376.0, 49936527360.0, 23787079680.0, 7051922944.0, 9142196224.0, 9748790272.0, 3319812864.0, 2783002624.0, 3333287680.0, 1974285952.0, 61073563648.0, 23916335104.0, 3902347008.0, 5904737792.0, 5904737792.0, 7679072768.0, 104155185152.0, 67303845888.0, 9242958848.0, 11397070848.0, 22049380352.0, 1826739322880.0, 159639027712.0, 31965216768.0, 5249129984.0, 1891344640.0, 1573129216.0, 1314357120.0, 4008590080.0, 8773196800.0, 20124471296.0, 16760728576.0, 16760728576.0, 9958882304.0, 5928898560.0, 4343258624.0, 1916079360.0, 939273664.0, 629265088.0, 481244416.0, 205151792.0, 137910896.0, 110538704.0, 168628448.0, 222643696.0, 253653280.0, 222960896.0, 139691040.0, 121091456.0, 72755192.0, 58658920.0, 56773184.0, 91250544.0, 117835960.0, 1534850048.0, 2316171264.0, 2258605824.0]
loss_rpn_box_reg: [0.0122, 0.023, 0.0206, 0.0176, 0.0175, 0.017, 0.0152, 0.0145, 0.0146, 0.015, 0.0117, 0.0117, 0.0042, 0.0093, 0.0107, 0.0102, 0.0086, 0.0113, 0.0099, 0.0085, 0.0086, 0.0102, 0.0093, 0.0096, 0.009, 0.0111, 0.0111, 0.0107, 0.0071, 0.007, 0.0071, 0.0076, 0.0084, 0.0084, 0.0075, 0.0075, 0.0094, 0.0058, 0.0069, 0.008, 0.006, 0.0069, 0.007, 0.0078, 0.0068, 0.0064, 0.0061, 0.0061, 0.005, 0.0061, 0.0066, 0.0077, 0.0059, 0.0045, 0.0049, 0.0048, 0.0044, 0.0044, 0.0058, 0.0066, 0.0178, 0.0063, 0.0059, 0.0058, 0.005, 0.0053, 0.0056, 0.0042, 0.0037, 0.0052, 0.0054, 0.0047, 0.0083, 4122.6558, 16300.8662, 17248.6602, 7573.0581, 4729.5298, 7850.1875, 24332.0605, 24332.0605, 6414.6733, 6141.377, 11858.5957, 1523679.875, 57047.1758, 88967.75, 113751.0234, 325364.8438, 1778727.375, 1167142.625, 3055371.75, 11142872.0, 12528507.0, 15887939.0, 21355252.0, 16677439488.0, 51280272.0, 12814747.0, 6924202.5, 5021109.5, 1700890.25, 862957.8125, 41252116.0, 41425260.0, 22989032.0, 47973796.0, 43746984.0, 26487380.0, 174136864.0, 352447072.0, 194540448.0, 120696296.0, 226160800.0, 63684192.0, 74702128.0, 1971091328.0, 5083697152.0, 44886159360.0, 34794504192.0, 943304640.0, 15580342272.0, 25458366464.0, 25458366464.0, 64111980544.0, 64023056384.0, 22132686848.0, 29597784064.0, 29597784064.0, 10248808448.0, 6831554560.0, 7227887616.0, 4142664192.0, 144436805632.0, 44608315392.0, 6446649344.0, 11398449152.0, 19815819264.0, 20694464512.0, 119358201856.0, 98220417024.0, 17584498688.0, 68699357184.0, 108964569088.0, 12136470282240.0, 510677483520.0, 174832304128.0, 13222363136.0, 4024046336.0, 4468099584.0, 2940316416.0, 8586808832.0, 20942501888.0, 46743859200.0, 43685576704.0, 52968079360.0, 15070398464.0, 8562162176.0, 8059048960.0, 4783551488.0, 2039989632.0, 1187719808.0, 456337568.0, 341187520.0, 286260928.0, 216146336.0, 462010880.0, 497745760.0, 289512544.0, 375960640.0, 297198336.0, 137349200.0, 146000160.0, 122740864.0, 125930776.0, 149373344.0, 215782768.0, 3676463616.0, 6298272256.0, 5892077056.0]
model_time: []
evaluator_time: []
total_time: []

# Example: Extracting the average recall for different thresholds
recall_values_SGDADAMN = metrics["AR_100"]  # Let's use AR_100 as an example for IoU thresholds plotting

# Example: Precision vs. Recall (assuming AP data correlates with precision directly at different recalls)
precision_values_SGDADAMN = metrics["AP"]  # Direct extraction for simplicity in this example

import matplotlib.pyplot as plt

# Sort the data by recall since the precision-recall curve expects this.
sorted_indices_ADAMNSGD = sorted(range(len(recall_values_ADAMNSGD)), key=lambda k: recall_values_ADAMNSGD[k])
precision_values_ADAMNSGD_sorted = [precision_values_ADAMNSGD[i] for i in sorted_indices_ADAMNSGD]
recall_values_ADAMNSGD_sorted = [recall_values_ADAMNSGD[i] for i in sorted_indices_ADAMNSGD]

# To ensure the plot fully spans, check starts and ends
if recall_values_ADAMNSGD_sorted[0] > 0:
    recall_values_ADAMNSGD_sorted.insert(0, 0)
    precision_values_ADAMNSGD_sorted.insert(0, precision_values_ADAMNSGD[0])

if recall_values_ADAMNSGD_sorted[-1] < 1:
    recall_values_ADAMNSGD_sorted.append(1)
    precision_values_ADAMNSGD_sorted.append(precision_values_ADAMNSGD[-1])

# Sort the data for SGD since the precision-recall curve expects this.
sorted_indices_SGD = sorted(range(len(recall_values_SGD)), key=lambda k: recall_values_SGD[k])
precision_values_SGD_sorted = [precision_values_SGD[i] for i in sorted_indices_SGD]
recall_values_SGD_sorted = [recall_values_SGD[i] for i in sorted_indices_SGD]

# To ensure the plot fully spans, check starts and ends
if recall_values_SGD_sorted[0] > 0:
    recall_values_SGD_sorted.insert(0, 0)
    precision_values_SGD_sorted.insert(0, precision_values_SGD[0])

if recall_values_SGD_sorted[-1] < 1:
    recall_values_SGD_sorted.append(1)
    precision_values_SGD_sorted.append(precision_values_SGD[-1])

# Sort the data for SGDADAMN since the precision-recall curve expects this.
sorted_indices_SGDADAMN = sorted(range(len(recall_values_SGDADAMN)), key=lambda k: recall_values_SGDADAMN[k])
precision_values_SGDADAMN_sorted = [precision_values_SGDADAMN[i] for i in sorted_indices_SGDADAMN]
recall_values_SGDADAMN_sorted = [recall_values_SGDADAMN[i] for i in sorted_indices_SGDADAMN]

# To ensure the plot fully spans, check starts and ends
if recall_values_SGDADAMN_sorted[0] > 0:
    recall_values_SGDADAMN_sorted.insert(0, 0)
    precision_values_SGDADAMN_sorted.insert(0, precision_values_SGDADAMN[0])

if recall_values_SGDADAMN_sorted[-1] < 1:
    recall_values_SGDADAMN_sorted.append(1)
    precision_values_SGDADAMN_sorted.append(precision_values_SGDADAMN[-1])

# Sort the data for Adam since the precision-recall curve expects this.
sorted_indices_Adam = sorted(range(len(recall_values_Adam)), key=lambda k: recall_values_Adam[k])
precision_values_Adam_sorted = [precision_values_Adam[i] for i in sorted_indices_Adam]
recall_values_Adam_sorted = [recall_values_Adam[i] for i in sorted_indices_Adam]

# To ensure the plot fully spans, check starts and ends
if recall_values_Adam_sorted[0] > 0:
    recall_values_Adam_sorted.insert(0, 0)
    precision_values_Adam_sorted.insert(0, precision_values_Adam[0])

if recall_values_Adam_sorted[-1] < 1:
    recall_values_Adam_sorted.append(1)
    precision_values_Adam_sorted.append(precision_values_Adam[-1])

# Sort the data for RMSprop since the precision-recall curve expects this.
sorted_indices_RMSprop = sorted(range(len(recall_values_RMSprop)), key=lambda k: recall_values_RMSprop[k])
precision_values_RMSprop_sorted = [precision_values_RMSprop[i] for i in sorted_indices_RMSprop]
recall_values_RMSprop_sorted = [recall_values_RMSprop[i] for i in sorted_indices_RMSprop]

# To ensure the plot fully spans, check starts and ends
if recall_values_RMSprop_sorted[0] > 0:
    recall_values_RMSprop_sorted.insert(0, 0)
    precision_values_RMSprop_sorted.insert(0, precision_values_RMSprop[0])

if recall_values_RMSprop_sorted[-1] < 1:
    recall_values_RMSprop_sorted.append(1)
    precision_values_RMSprop_sorted.append(precision_values_RMSprop[-1])

# Sort the data for Adelta since the precision-recall curve expects this.
sorted_indices_Adelta = sorted(range(len(recall_values_Adelta)), key=lambda k: recall_values_Adelta[k])
precision_values_Adelta_sorted = [precision_values_Adelta[i] for i in sorted_indices_Adelta]
recall_values_Adelta_sorted = [recall_values_Adelta[i] for i in sorted_indices_Adelta]

# To ensure the plot fully spans, check starts and ends
if recall_values_Adelta_sorted[0] > 0:
    recall_values_Adelta_sorted.insert(0, 0)
    precision_values_Adelta_sorted.insert(0, precision_values_Adelta[0])

if recall_values_Adelta_sorted[-1] < 1:
    recall_values_Adelta_sorted.append(1)
    precision_values_Adelta_sorted.append(precision_values_Adelta[-1])

# Create the step plot for the precision-recall curve
plt.figure(figsize=(10, 5))

# Plot for AdamSGD
plt.step(recall_values_ADAMNSGD_sorted, precision_values_ADAMNSGD_sorted, where='post', color='purple', linewidth=2.5, label='Precision vs. Recall (AdamSGD)')

# Plot for SGD
plt.step(recall_values_SGD_sorted, precision_values_SGD_sorted, where='post', color='blue', linewidth=2.5, label='Precision vs. Recall (SGD)')

# Plot for SGDADAMN
plt.step(recall_values_SGDADAMN_sorted, precision_values_SGDADAMN_sorted, where='post', color='green', linewidth=2.5, label='Precision vs. Recall (SGDADAMN)')

# Plot for Adam
plt.step(recall_values_Adam_sorted, precision_values_Adam_sorted, where='post', color='orange', linewidth=2.5, label='Precision vs. Recall (Adam)')

# Plot for RMSprop
plt.step(recall_values_RMSprop_sorted, precision_values_RMSprop_sorted, where='post', color='red', linewidth=2.5, label='Precision vs. Recall (RMSprop)')

# Plot for Adelta
plt.step(recall_values_Adelta_sorted, precision_values_Adelta_sorted, where='post', color='cyan', linewidth=2.5, label='Precision vs. Recall (Adelta)')

# Customize the plot
plt.xlabel('Recall')
plt.ylabel('Precision')

# Title for AdamSGD
plt.text(0.15, 1.05, ' --AdamSGD', color='purple', fontsize=12, ha='left', va='bottom', transform=plt.gca().transAxes)

# Title for SGD
plt.text(0.30, 1.05, ' --SGD', color='blue', fontsize=12, ha='left', va='bottom', transform=plt.gca().transAxes)

# Title for SGDADAMN
plt.text(0.38, 1.05, ' --SGDADAMN', color='green', fontsize=12, ha='left', va='bottom', transform=plt.gca().transAxes)

# Title for Adam
plt.text(0.55, 1.05, ' --Adam', color='orange', fontsize=12, ha='left', va='bottom', transform=plt.gca().transAxes)

# Title for RMSprop
plt.text(0.65, 1.05, ' --RMSprop', color='red', fontsize=12, ha='left', va='bottom', transform=plt.gca().transAxes)

# Title for Adelta
plt.text(0.79, 1.05, ' --Adelta', color='cyan', fontsize=12, ha='left', va='bottom', transform=plt.gca().transAxes)


plt.xlim(0, 1)
plt.ylim(0, 1.2)
plt.grid(True)

# Draw a horizontal line at y=1
plt.axhline(y=1, color='red', linestyle='-', linewidth=3.5, label='Max Precision')

# Enhance the legend to include stylistic references
plt.legend(title='Legend', bbox_to_anchor=(1.05, 1), loc='upper left')

plt.show()

import matplotlib.pyplot as plt

# Define the IoU thresholds from 0 to 1 with an increment of 0.2
iou_thresholds = [i / 5 for i in range(6)]

# Your provided recall values. Assuming each array has more than 6 elements, slice to the first 6.
recall_values_ADAMNSGD = recall_values_ADAMNSGD[:6]  # Slice to the first 6 values
recall_values_SGD = recall_values_SGD[:6]            # Slice to the first 6 values
recall_values_SGDADAMN = recall_values_SGDADAMN[:6]  # Slice to the first 6 values
recall_values_Adam = recall_values_Adam[:6]          # Slice to the first 6 values
recall_values_RMSprop = recall_values_RMSprop[:6]    # Slice to the first 6 values
recall_values_Adelta = recall_values_Adelta[:6]      # Slice to the first 6 values

plt.figure(figsize=(10, 5))

# Plot each optimizer with points marked by 'o'
plt.plot(iou_thresholds, recall_values_ADAMNSGD, marker='o', linestyle='-', color='cyan', label='AdamSGD')
plt.plot(iou_thresholds, recall_values_SGD, marker='o', linestyle='-', color='purple', label='SGD')
plt.plot(iou_thresholds, recall_values_SGDADAMN, marker='o', linestyle='-', color='blue', label='SGDADAMN')
plt.plot(iou_thresholds, recall_values_Adam, marker='o', linestyle='-', color='green', label='Adam')
plt.plot(iou_thresholds, recall_values_RMSprop, marker='o', linestyle='-', color='orange', label='RMSprop')
plt.plot(iou_thresholds, recall_values_Adelta, marker='o', linestyle='-', color='red', label='Adelta')

# Title for AdamSGD
plt.text(0.15, 1.05, ' --AdamSGD', color='cyan', fontsize=12, ha='left', va='bottom', transform=plt.gca().transAxes)

# Title for SGD
plt.text(0.30, 1.05, ' --SGD', color='purple', fontsize=12, ha='left', va='bottom', transform=plt.gca().transAxes)

# Title for SGDADAMN
plt.text(0.38, 1.05, ' --SGDADAMN', color='blue', fontsize=12, ha='left', va='bottom', transform=plt.gca().transAxes)

# Title for Adam
plt.text(0.55, 1.05, ' --Adam', color='green', fontsize=12, ha='left', va='bottom', transform=plt.gca().transAxes)

# Title for RMSprop
plt.text(0.65, 1.05, ' --RMSprop', color='orange', fontsize=12, ha='left', va='bottom', transform=plt.gca().transAxes)

# Title for Adelta
plt.text(0.79, 1.05, ' --Adelta', color='red', fontsize=12, ha='left', va='bottom', transform=plt.gca().transAxes)

# Customize the plot
plt.xlabel('IoU Threshold')
plt.ylabel('Recall')
plt.xticks(iou_thresholds)
plt.xlim(0, 1)
plt.ylim(min(min(recall_values_ADAMNSGD, recall_values_SGD, recall_values_SGDADAMN, recall_values_Adam, recall_values_RMSprop, recall_values_Adelta)) - 0.05,
         max(max(recall_values_ADAMNSGD, recall_values_SGD, recall_values_SGDADAMN, recall_values_Adam, recall_values_RMSprop, recall_values_Adelta)) + 0.05)
plt.grid(True)
plt.legend(title='Optimizer', loc='upper right')

plt.show()

import pickle


# Define the file path where you want to save the model
Filename = "/content/drive/MyDrive/dataset/FRCNN2adamn.pkl"
# Save the Modle to file in the current working directory
with open(Filename, 'wb') as file:
    pickle.dump(model, file)
# Load the Model back from file
with open(Filename, 'rb') as file:
    model = pickle.load(file)
model

FasterRCNN(
  (transform): GeneralizedRCNNTransform(
      Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
      Resize(min_size=(800,), max_size=1333, mode='bilinear')
  )
  (backbone): BackboneWithFPN(
    (body): IntermediateLayerGetter(
      (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)
      (bn1): FrozenBatchNorm2d(64, eps=0.0)
      (relu): ReLU(inplace=True)
      (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)
      (layer1): Sequential(
        (0): Bottleneck(
          (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(64, eps=0.0)
          (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(64, eps=0.0)
          (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(256, eps=0.0)
          (relu): ReLU(inplace=True)
          (downsample): Sequential(
            (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
            (1): FrozenBatchNorm2d(256, eps=0.0)
          )
        )
        (1): Bottleneck(
          (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(64, eps=0.0)
          (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(64, eps=0.0)
          (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(256, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (2): Bottleneck(
          (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(64, eps=0.0)
          (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(64, eps=0.0)
          (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(256, eps=0.0)
          (relu): ReLU(inplace=True)
        )
      )
      (layer2): Sequential(
        (0): Bottleneck(
          (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(128, eps=0.0)
          (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(128, eps=0.0)
          (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(512, eps=0.0)
          (relu): ReLU(inplace=True)
          (downsample): Sequential(
            (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)
            (1): FrozenBatchNorm2d(512, eps=0.0)
          )
        )
        (1): Bottleneck(
          (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(128, eps=0.0)
          (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(128, eps=0.0)
          (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(512, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (2): Bottleneck(
          (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(128, eps=0.0)
          (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(128, eps=0.0)
          (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(512, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (3): Bottleneck(
          (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(128, eps=0.0)
          (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(128, eps=0.0)
          (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(512, eps=0.0)
          (relu): ReLU(inplace=True)
        )
      )
      (layer3): Sequential(
        (0): Bottleneck(
          (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
          (downsample): Sequential(
            (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False)
            (1): FrozenBatchNorm2d(1024, eps=0.0)
          )
        )
        (1): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (2): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (3): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (4): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (5): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256, eps=0.0)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256, eps=0.0)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(1024, eps=0.0)
          (relu): ReLU(inplace=True)
        )
      )
      (layer4): Sequential(
        (0): Bottleneck(
          (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(512, eps=0.0)
          (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(512, eps=0.0)
          (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(2048, eps=0.0)
          (relu): ReLU(inplace=True)
          (downsample): Sequential(
            (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False)
            (1): FrozenBatchNorm2d(2048, eps=0.0)
          )
        )
        (1): Bottleneck(
          (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(512, eps=0.0)
          (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(512, eps=0.0)
          (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(2048, eps=0.0)
          (relu): ReLU(inplace=True)
        )
        (2): Bottleneck(
          (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(512, eps=0.0)
          (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(512, eps=0.0)
          (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): FrozenBatchNorm2d(2048, eps=0.0)
          (relu): ReLU(inplace=True)
        )
      )
    )
    (fpn): FeaturePyramidNetwork(
      (inner_blocks): ModuleList(
        (0): Conv2dNormActivation(
          (0): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1))
        )
        (1): Conv2dNormActivation(
          (0): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1))
        )
        (2): Conv2dNormActivation(
          (0): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1))
        )
        (3): Conv2dNormActivation(
          (0): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1))
        )
      )
      (layer_blocks): ModuleList(
        (0-3): 4 x Conv2dNormActivation(
          (0): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
        )
      )
      (extra_blocks): LastLevelMaxPool()
    )
  )
  (rpn): RegionProposalNetwork(
    (anchor_generator): AnchorGenerator()
    (head): RPNHead(
      (conv): Sequential(
        (0): Conv2dNormActivation(
          (0): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
          (1): ReLU(inplace=True)
        )
      )
      (cls_logits): Conv2d(256, 3, kernel_size=(1, 1), stride=(1, 1))
      (bbox_pred): Conv2d(256, 12, kernel_size=(1, 1), stride=(1, 1))
    )
  )
  (roi_heads): RoIHeads(
    (box_roi_pool): MultiScaleRoIAlign(featmap_names=['0', '1', '2', '3'], output_size=(7, 7), sampling_ratio=2)
    (box_head): TwoMLPHead(
      (fc6): Linear(in_features=12544, out_features=1024, bias=True)
      (fc7): Linear(in_features=1024, out_features=1024, bias=True)
    )
    (box_predictor): FastRCNNPredictor(
      (cls_score): Linear(in_features=1024, out_features=11, bias=True)
      (bbox_pred): Linear(in_features=1024, out_features=44, bias=True)
    )
  )
)

# function to convert a torchtensor back to PIL image
def torch_to_pil(img):
    return torchtrans.ToPILImage()(img).convert('RGB')

# pick one image from the test set
img, target = dataset_test[30]
# put the model in evaluation mode
model.eval()
with torch.no_grad():
    prediction = model([img.to(device)])[0]

print('predicted #boxes: ', prediction['labels'])
print('real #boxes: ', target['labels'])

predicted #boxes:  tensor([10, 10, 10, 10, 10, 10, 10,  1, 10], device='cuda:0')
real #boxes:  tensor([10, 10, 10, 10, 10, 10])

import numpy as np
import matplotlib.pyplot as plt
from PIL import Image

def plot_img_bbox(img, target):
    # Convert PIL Image to NumPy array
    img_array = np.array(img)

    # Permute dimensions if it's a PyTorch tensor
    if isinstance(img_array, torch.Tensor):
        img_array = img_array.permute(1, 2, 0)

    # Plot the image and bounding boxes
    fig, a = plt.subplots(1,1)
    fig.set_size_inches(5,5)
    a.imshow(img_array)
    for box in target['boxes']:
        x, y, w, h = box
        rect = plt.Rectangle((x, y), w, h, fill=False, edgecolor='red', linewidth=2)
        a.add_patch(rect)
    plt.show()

print('EXPECTED OUTPUT')
plot_img_bbox(torch_to_pil(img), target)

EXPECTED OUTPUT

import torch
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.patches as patches

def plot_img_bbox(img, target):
    fig, a = plt.subplots(1, 1)
    fig.set_size_inches(5, 5)
    a.imshow(img)
#     print(target['boxes'])
    for box in target['boxes']:
        x, y, w, h = box.cpu().numpy()  # Move the tensor to CPU and convert to NumPy array
        #print("-------x------------")
        #print(x)
        #print("-------y------------")
        #print(y)
        #print("-------w------------")
        #print(w)
        #print("-------h------------")
        #print(h)
        width, height  = w-x, h-y
        rect = patches.Rectangle((x, y), width, height, linewidth=2, edgecolor='r', facecolor='none')
        a.add_patch(rect)
    plt.show()

print('MODEL OUTPUT')
plot_img_bbox(torch_to_pil(img), prediction)

MODEL OUTPUT

# the function takes the original prediction and the iou threshold.

def apply_nms(orig_prediction, iou_thresh=0.3):

    # torchvision returns the indices of the bboxes to keep
    keep = torchvision.ops.nms(orig_prediction['boxes'], orig_prediction['scores'], iou_thresh)

    final_prediction = orig_prediction
    final_prediction['boxes'] = final_prediction['boxes'][keep]
    final_prediction['scores'] = final_prediction['scores'][keep]
    final_prediction['labels'] = final_prediction['labels'][keep]

    return final_prediction

nms_prediction = apply_nms(prediction, iou_thresh=0.2)
print('NMS APPLIED MODEL OUTPUT')
plot_img_bbox(torch_to_pil(img), nms_prediction)

NMS APPLIED MODEL OUTPUT